General rights Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights.

Users may download and print one copy of any publication from the public portal for the purpose of private study or research.

You may not further distribute the material or use it for any profit-making activity or commercial gain

You may freely distribute the URL identifying the publication in the public portal If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from orbit.dtu.dk on: Sep 11, 2021

Physiology and metabolism of Yarrowia lipolytica for the utilization of alternativecarbon substrates

Lubuta, Patrice Jeremie Keta

Publication date:2018

Document VersionPublisher's PDF, also known as Version of record

Link back to DTU Orbit

Citation (APA):Lubuta, P. J. K. (2018). Physiology and metabolism of Yarrowia lipolytica for the utilization of alternative carbonsubstrates. Technical University of Denmark.

Physiology and metabolism of Yarrowia lipolytica

for the utilization of alternative carbon substrates

Patrice Lubuta

Ph.D. Thesis

December 2018

Physiology and metabolism of Yarrowia lipolytica for the

utilization of alternative carbon substrates

Ph.D. Thesis

Patrice Lubuta

Department of Bioengineering and Biomedicine

December 2018

Technical University of Denmark

2800 Kgs. Lyngby

Supervisors

Associate Professor, Ph.D. Christopher T. Workman

Associate Professor, Ph.D. Mhairi Workman

Contents Abstract ....................................................................................................................................................... 1

Danske resume ............................................................................................................................................ 2

Manuscripts ................................................................................................................................................. 3

Acknowledgements ...................................................................................................................................... 4

Part I: Introductory Chapters ........................................................................................................................ 6

Chapter 1: Industrial Biotechnology and the bioeconomic challenge .................................................................7

1.1 A brief history of Industrial Biotechnology .................................................................................................7

1.2 The bioeconomy vision ...............................................................................................................................9

1.3 Microbial cell factories ............................................................................................................................ 11

1.4 Glycerol and lignocellulosic sugars: alternative substrates for microbial fermentation processes ........ 14

Chapter 2: Yarrowia lipolytica and review of its carbon metabolism ............................................................... 16

2.1 Y. lipolytica: A promising host for novel biotechnological applications .................................................. 16

2.2 Glycerol metabolism, transport and regulation ...................................................................................... 18

2.3 Pentose metabolism ................................................................................................................................ 28

2.4 Sugar transport in Y. lipolytica ................................................................................................................. 31

Chapter 3: Methods of cell factory characterization and analysis .................................................................... 33

3.1 Quantitative physiology ........................................................................................................................... 33

3.2 Genomics ................................................................................................................................................. 35

References of Part I ........................................................................................................................................... 41

Part II: Manuscripts .................................................................................................................................... 51

Manuscript 1: Physiological comparison of Yarrowia lipolytica strains reveals differences in the utilization of

sugars and glycerol ............................................................................................................................................ 52

Abstract ......................................................................................................................................................... 53

Introduction ................................................................................................................................................... 54

Results ........................................................................................................................................................... 55

Discussion ...................................................................................................................................................... 61

Materials and methods ................................................................................................................................. 65

References ..................................................................................................................................................... 69

Supplemental material .................................................................................................................................. 74

Manuscript 2: Draft Genome Sequences of Yarrowia lipolytica Strains H222, IBT 446 and W29 .................... 80

Abstract ......................................................................................................................................................... 81

Introduction ................................................................................................................................................... 81

Results and Discussion ................................................................................................................................... 81

References ..................................................................................................................................................... 83

Manuscript 3: Genome-wide expression analysis of Yarrowia lipolytica strains varying in the utilization of

glucose and glycerol .......................................................................................................................................... 85

Abstract ......................................................................................................................................................... 86

Introduction ................................................................................................................................................... 87

Materials and Methods ................................................................................................................................. 89

Results and Discussion ................................................................................................................................... 93

Supplemental material ................................................................................................................................ 125

Conclusions and Perspectives ................................................................................................................... 131

1

Abstract Efforts to valorize alternative carbon feedstocks from lignocellulosic hydrolysis or the biodiesel by-product

glycerol have motivated investigations into new cell-factory hosts. The non-conventional yeast species

Yarrowia lipolytica has attracted attention in recent years as a promising candidate for these novel and

sustainable biotechnological applications. In this Ph.D. thesis, we analyzed the physiology, genetics and

metabolism of Y. lipolytica for the usage of alternative carbon sources by methods of quantitative physiology

and genomics.

We benchmarked the cellular performance of the three Y. lipolytica strains IBT 446, W29 and H222 on glucose,

xylose, arabinose and glycerol using single and mixed substrate fermentations in controlled bioreactors. Glycerol

was found to be the preferred carbon source for all three strains, leading to the highest growth rates and the

production of sugar alcohols. Inter-strain variations were detected and, in particular, IBT 446 was found to differ

in several characteristics from the commonly used strains W29 and H222. IBT 446, originally isolated from Danish

feta cheese, possessed beneficial characteristics like the absence of hyphal growth, which usually causes

problems in industrial fermentations. Since physiological differences were observed, we sequenced the genomes

of the three Y. lipolytica strains.

All strains showed a characteristic sequential substrate utilization in mixed carbon fermentations. Interestingly,

it was observed that the presence of glycerol can prevent the consumption of glucose and that this suppression

is further strain dependent: IBT 446 exhibited a strong sequential utilization of glycerol and glucose, whereas

W29 co-consumed the two substrates. This indicated so far unknown carbon regulation mechanisms, which are

converse to well-described carbon repression systems (e.g. in S. cerevisiae or E. coli) ensuring the prioritized use

of glucose. RNAseq analysis was performed in order to investigate the influence of glycerol on the gene

expression. We could show that genes encoding several transporters and metabolic enzymes were expressed

significantly higher in W29. Further, we found strain-specific carbon responses and that several differentially

expressed genes encode proteins related to signal transduction and transcriptional regulation, e.g. S. cerevisiae

orthologs RME1, STE4, STE6, SST2, GPA2 and AZF1.

2

Danske resume Bestræbelser på at valorisere alternative kulstofkilder fra lignocellulose hydrolyse eller biodiesel biproduktet

glycerol har drevet undersøgelser af nye cellefabrik-værter. Den ikke konventionelle gærart Yarrowia lipolytica

har i de senere år tiltrukket opmærksomhed som en lovende kandidat indenfor disse nye og bæredygtige

bioteknologiske anvendelser. I denne ph.d. afhandling har vi analyseret fysiologien, genetikken og metabolismen

af Y. lipolytica, ved at bruge metoder såsom kvantitativ fysiologi og genetik, med henblik på at udnytte alternative

kulstofkilder.

Vi har sammenlignet den cellulære ydeevne af tre Y. lipolytica stammer (IBT 446, W29 and H222) ved

fermentering af enkelt kulstof substrat som glukose, xylose, arabinose og glycerol i kontrollerede bioreaktorer,

såvel som ved blandet substrat fermentering. Den foretrukne kulstofkilde for alle tre stammer er glycerol, hvilket

ledte til de højeste vækstrater samt produktion af sukkeralkoholer. Variationer mellem stammerne blev

observeret, og specielt IBT 446 adskiller sig i flere egenskaber sammenlignet med de mest anvendte stammer

W29 and H222. IBT 446, som oprindeligt er isoleret fra dansk fetaost, besidder gavnlige egenskaber som fraværet

af hyfevækst, hvilket ofte skaber problemer i industrielle fermenteringer. Eftersom fysiologiske forskelle blev

observeret, blev genomerne af de tre Y. lipolytica stammer sekventeret.

Alle stammer viste en karakteristisk sekventiel substratudnyttelse i blandede kulstof-fermenteringer.

Tilstedeværelsen af glycerol kunne overraskende nok forhindre forbrug af glukose, derudover kunne det påvises,

at denne undertrykkelse af glukoseforbrug er stamme afhængig: IBT 446 udviste en stærk sekventiel udnyttelse

af glycerol efterfulgt af glukose, hvorimod W29 forbrugte de to substrater samtidig. RNAseq analyse blev udført

på prøver fra chemostat kultiveringer groet på glukose, glycerol og en glukose-glycerol blanding for at undersøge

indflydelsen af glycerol på gen-ekspressionen. Vi kunne derved vise at gener, der koder for flere transport og

metaboliske enzymer, blev udtrykt signifikant højere in W29. Derudover fandt vi stamme-specifikke kulstof

responser samt at flere forskelligt udtrykte gener koder for proteiner, som er relateret til signaltransduktion og

transkriptionel regulering, eksempelvis S. cerevisiae orthologer RME1, STE4, STE6, SST2, GPA2 and AZF1.

3

Manuscripts

The work conducted in this Ph.D. thesis formed the basis for the following manuscripts:

Patrice Lubuta, Christopher T. Workman and Mhairi Workman,

Physiological comparison of Yarrowia lipolytica strains reveals differences in the

utilization of sugars and glycerol.

Submitted to Applied and Environmental Microbiology (AEM), 2018

Patrice Lubuta, Mhairi Workman and Christopher T. Workman,

Draft Genome Sequences of Yarrowia lipolytica Strains H222, IBT 446 and W29.

Submitted Microbiology Resource Announcements, 2018

Patrice Lubuta, Mhairi Workman, Eduard Kerkhoven & Christopher T. Workman,

Genome-wide expression analysis of Yarrowia lipolytica strains varying in the

utilization of glucose and glycerol.

Submitted Genes, Genomes, Genetics (G3), 2018

4

Acknowledgements This PhD thesis was conducted at the Technical University of Denmark, Department of Bioengineering and

Biomedicine between January 2015 and December 2018. The study was funded by a PhD stipend from the

Technical University of Denmark. The main supervisor of this study was Christopher T. Workman who took over

the supervision from the original main supervisor Mhairi Workman.

First of all, I would like to express my deepest gratitude to Christopher T. Workman who took over the main

supervision of this PhD project. I am grateful for all the supervision and great guidance. I am especially thankful

that by working together with Chris, I had the opportunity to learn a real interdisciplinary approach to tackle

biological questions. Through my bachelor and master studies I could gain knowledge in the fields of molecular

biology, microbiology and genetics. Coming to DTU I was able to broaden my knowledge in the area of

fermentation technology and quantitative physiology. After the supervisor change, the project shifted into the

area of genomics and sequencing based technologies. By working together with Chris, I learned how powerful

the use of computer science in biological research is. Learning how to run bioinformatics software on the Linux

command line and using the R programming language for statistical analysis was often hard and also frustrating

but gave me an inestimable worth for my scientific career.

I also would like to express my greatest gratitude to my original main supervisor Mhairi Workman who put trust

in me and offered me the opportunity to conduct a PhD here at DTU. It was a great experience to work on DTU´s

fermentation platform. I am grateful for all of the guidance of Mhairi also after leaving the University.

Next, I like to express gratitude to Eduard Kerkhoven who made it possible to conduct an external stay at the

Chalmers University of Technology in Gothenburg, Sweden. He was always available and a quick responder to

the many questions I had concerning Y. lipolytica, the R programming language and data analysis.

5

None of the work would have been possible without huge efforts of Tina Johansen, Martin Nielsen, Alexander

Rosenkjær, and Andreas Heidemann. Thanks for the great technical assistance and problem solving related to

the fermentation equipment and HPLC systems.

I would like to express special thanks to all my colleagues and co-workers at the department. Long nights in the

fermentation lab wouldn't be half as much fun without the good conversations and cooking sessions in the

kitchen of building 223. There are too many people to mention of course but a special thanks goes to Ferdinand

Kirchner, Julian Brandl, Sietske Grijseels, Elise de Reus, Cyrielle Calmels, Milica Randjelovic, Sebastian Theobald,

Anantha Peramuna and Hansol Bae (the latter two for the fun nights at DTU Kælderbaren). Special thanks goes

to Inge Kjærbølling for translating the thesis abstract into Danish.

Finally, inestimable thanks goes to my family and friends in Germany, who were endlessly patient with me and

not having me around for a couple of years. Without the support, it would not be possible to accomplish this

PhD thesis.

My most sincere apologies goes to all I have forgotten to mention in this text.

Patrice Lubuta, December 2018

6

Part I: Introductory Chapters

7

Chapter 1: Industrial Biotechnology and the

bioeconomic challenge

The following work is settled in the field of Industrial Biotechnology. Industrial Biotechnology can be defined as

any technological application that uses biological systems, living organisms, or derivatives thereof (e.g. enzymes),

to make or modify products or processes for a specific use in industry (UN Convention on Biological Diversity).

This field is also known as White biotechnology, and is distinguished from other areas such as Red (health-related

applications) and Green (agricultural) biotechnology. Biotechnological products range from fuel and chemicals

over food ingredients to pharmaceuticals.

1.1 A brief history of Industrial Biotechnology

Even though being considered as a key technology of the 21th century, biotechnological principles have been

used by human kind since ancient time. Without being skilled biotechnologists, people from old Mesopotamia,

Egypt, China and India used microorganisms for the production of beer, bread and wine. Industrial biotechnology

relies largely on a biochemical process called fermentation, which is originally defined as the cellular

consumption of sugar in the absence of oxygen. In a broader biotechnological context, fermentation is regarded

as any microbial process in which specific substrates are converted into desired products. Early milestones on

the way to modern Industrial Biotechnology were the production of glycerol from sugars by the yeast

Saccharomyces cerevisiae during the First World War, and the production of acetone and butanol by the

bacterium Clostridium acetobutylicum during the Second World War (1). An exceedingly important role in the

evolution of modern Industrial Biotechnology was the discovery of penicillin in the late 1920s. The following high

demand for antibiotics accelerated the development of various biotechnological methods such as sterile

cultivation techniques, the worldwide search for fungal producers of natural products and optimization

strategies based on mutagenesis and screening (1). In the 1960s, single-cell protein (SCP) produced from

petroleum or natural gas was considered as a solution for impending food shortages in the increasing population

of third world countries, however, SCP became unfeasible due to improvements in other sectors. Several

discoveries in the late 1950s and early 1960s allowed the overproduction of amino acids in bacteria, especially

in Corynebacterium glutamicum. Mutants hindered in the enzymatic degradation of the desired amino acid had

been isolated, which enabled the production of amino acids which occurred only in insufficient amounts in plant

8

proteins. Since the 1970s, against the background of global warming and oil shortages, a sustainable energy

supply as well as environmental protection became increasingly important. The production of biofuels grew

rapidly mainly in Brazil and the United States. However, as discussed below, producing bioethanol from

renewable feedstocks like sugarcane or corn, comes along with several disadvantages. Modern Industrial

Biotechnology is massively influenced by various innovations made in other disciplines. Since the 1980s,

recombinant DNA technology enabled the production of therapeutic proteins (e.g. insulin, monoclonal

antibodies), which became the main products of the biopharmaceutical industry (1). Advances in massive parallel

genome sequencing and computer science have expanded the traditional in vivo and in vitro methodology with

in silico approaches such as bioinformatics or systems biology. Nowadays, the biotechnological sector became a

billion dollar marked and further growth is expected. Biotechnologically produced goods range from low to high

value and former ones are usually produced in high amounts (bulk products), while the latter are usually

produced only in small quantities (Figure 1) (2, 3). In the next section, it is discussed how the latest developments

in Industrial Biotechnology have the potential to revolutionize the way we produce our future transportation

fuels and chemicals.

Figure 1: Examples for high and low value-added Biotech products and their corresponding production volume. The figure was taken from Hong & Nielsen (2012).

9

1.2 The bioeconomy vision

The evolution of industrialization in the 20th century, led to a close connection between the petrochemical

industry and the production of society´s everyday consumer goods (2). Simple building blocks derived from fossil

resources are converted into various products such as transportation fuels, polymers, solvents, textiles,

pharmaceuticals, flavors and nutrients. Over a century, the processes and corresponding infrastructure have

been drastically improved so that modern ways of production are highly optimized, efficient and cheap. Today’s

industry is, therefore, also referred to as the petroleum based industry (4). However, deriving products from

fossil resources has several drawbacks. On one side, resources are limited and a reduced oil supply will increase

feedstock prices, even though the exploitation of new deposits lead to medium-term reductions of the price (5).

On the other side, there are concerns about global warming and environmental pollution caused by the

petroleum based chemical industry. These factors, have driven the search for alternative, environmentally

friendlier ways of production including the use of renewable feedstocks. In principle, biotechnological production

processes have the potential to overcome both of these problems, provided that it becomes possible to produce

fuels and chemicals from biomass efficiently. The use of renewable feedstocks allows the establishment of a

circular industry, with a reduced necessity for finite resources and, therefore, circumventing a one-way street of

carbon flux (Figure 2). Additionally, enzyme based biosynthesis takes place under environmentally friendlier

conditions, with the less use of toxic solvents. Since enzyme-based catalysis is more chemo-, regio-, and

stereoselective also the production of novel compounds becomes possible. In contrast to the petroleum-based

industry, the European Union had defined the bio-based industry, which is also known as the bioeconomy, as the

production of renewable biological resources and the conversion of these resources and waste streams into value

added products (6).

10

Figure 2: The bioeconomy vision. The use of renewable plant-based feedstocks in biotechnological processes allows to produce bio-based products. Degradation of these products releases CO2 which fixed again by photosynthesis. This enables circular economy. The figure was taken from Campbell, Xia, & Nielsen (2017).

The transition to a bio-based economy, however, is highly challenging and demands the cooperation of various

scientific disciplines with each other, but also with the politics and industry. Generally, nature provides a

comprehensive toolset in its biodiversity with a nearly infinite number of enzymes and metabolic pathways,

allowing the degradation of biomass and the synthesis of an enormous amount of (bio)organic molecules. These

molecules and compounds can replace a large number of fossil derived chemicals. For a long time raising this

treasure from nature was limited to the usage cultivatable microorganisms and their endogenous metabolic

machinery. Being dependent on only native production organisms, biotechnological engineers were faced to

various problems such as low titer, rates and yields (TRY), stress sensitivity or generally slow growth

characteristics (2). Optimization strategies were limited to the selection of other organism, untargeted

mutagenesis and screening approaches and process engineering. The vision of bioeconomy is therefore not

conceivable without a breakthrough technological progress: As mentioned above, the upcoming recombinant

DNA technology enabled to transfer genes from one organism to another, allowing to cross the natural species

boundaries. The possibility to express genes heterologously, to overexpress and to knock out certain genes was

the birth of metabolic engineering. This discipline strives to design industrial microorganisms rationally. Latest

11

developments in systems and synthetic biology expand possibilities even more. Enzymes and metabolic

pathways from former uncultivatable microorganisms, plants or higher organisms became thus accessible (2).

The technological breakthroughs made biotechnological production processes attractive for traditional fuel and

chemical companies. Today, various examples exists, in which these companies switched from a classical

chemical to a chemical-biotechnological or a solely biotechnological process (3). To enable the transition from

an oil-based to a bio-based production the biorefinery concept is crucial: In analogy to an oil refinery, in which

crude oil is transformed to more useful products (e.g. gasoline, naphtha, diesel, lubricating oils, etc.) the

biorefinery uses plant-based raw materials, which are not directly usable in the fermentation process. The

biomass gets treated and converted to fermentable sugar monomers. This process is already well established for

starch-based feedstocks, but still faces challenges when applied to lignocellulosic biomass (3).

1.3 Microbial cell factories

Central for any biotechnological application is the fermentation microorganism. An efficient fermentation

organism can be seen as a small biological factory by itself, which takes up specific substrates and produces

certain products, by-products and its own biomass (Figure 3). Because of the analogy to a miniaturized factory,

the fermentation organism is commonly called a cell factory. Traditionally, several organism have been used in

industry, comprising bacterial species (e.g. Corynebacterium glutamicum, Escherichia coli and Bacillus subtilis),

fungal species (e.g. Saccharomyces cerevisiae and Aspergillus spp.) but also cells from higher organisms (e.g.

Chinese Hamster Ovary cells) (2). These platform organisms have been extensively developed over the years

resulting in beneficial properties such as a high capacity to produce certain products, the tolerance to

fermentation inhibitors or significant knowledge of the physiology and genetics (8).

12

Figure 3: A microbial cell factory. In analogy to an industrial factory, a microbial production organism can be seen as a miniaturized factory, which takes up certain substrates and generates different products and by-products.

A biotechnological process can be roughly divided into three distinct phases: 1.) upstream processing (medium

preparation, sterilization, and inoculum preparation), 2.) microbial fermentation (generation of biomass and the

desired product) and 3.) downstream processing (product recovery, waste treatment). Process optimization

takes place in every phase, but the optimization of the fermentation organism, which is a biological system, is

especially challenging, time consuming and costly. Usually, the development starts with a proof-of-principle

strain which is able to produce the desired product in small quantities but lacks the economic requirements

(Figure 4A) (3). For a long time, mainly untargeted mutagenesis and screening approaches have been used to

optimize the fermentation organism.

With the upcoming of system wide analysis methods and metabolic engineering, it became possible to expand

the former random approaches by rationale strain engineering concepts. The interplay between in depth

biological knowledge and targeted genetic modifications transformed the task of cell factory optimization into

an information driven and iterative process (7). This process can be separated into specific phases, and is known

as the cell factory design cycle or the Design-Build-Test-Learn cycle (Figure 4B): Information about the cellular

performance of a given strain are gained by quantitative physiology methods (characterization or test phase).

Additional cellular and metabolic data is provided by omics-technologies such as genomics, transcriptomics,

proteomics or metabolomics (analysis phase). Cell factory characterization and analysis methods are discussed

in Chapter 3. In the following, the gained knowledge is used to design bioprocesses or to plan specific genetic

alterations (design phase). Finally, the tailor made strains are constructed using genetic engineering

13

(construction, synthesis or built phase). After one cycle, the constructed strains can be tested in the next round

of characterization.

Figure 4: Concept of modern cell factory development. (A): Modern strain optimization methods use in-depth biological information combined with targeted genetic modifications to reduce the time and costs compared to older methods. The figure is taken from Campbell et al. (2017). (B): These methods can be applied in an iterative design-built-test-learn cycle enabling rationale cell engineering. Methods of cell factory characterization and analysis are discussed in Chapter 3.

14

1.4 Glycerol and lignocellulosic sugars: alternative substrates for microbial

fermentation processes

Glucose or starch derived from plants such as sugar beet, sugarcane, corn or wheat are examples for easy-to-use

renewable raw materials for biotechnological processes. Due to the rapid growth of fermentative applications,

especially the biofuel sector, the demand for sugars has risen sharply. However, since these raw materials can

also be used for the food and feed production, their use as fermentation substrates is highly controversial (“food

versus fuel debate”). Renewable resources that compete with the food and feed production are referred to as

first-generation substrates. Therefore, second-generation substrates are needed, which do not have these

disadvantages (9). Several alternative and sustainable feedstocks are available and in the following lignocellulosic

sugars and glycerol are presented.

Lignocellulosic biomass is the most abundant renewable biological resource on earth. Derived from agricultural

and forestry waste or nonfood crops, it is predestined for second-generation biorefinery applications. The

composition of lignocellulose depends on the used plant biomass but contain in average 35–50% cellulose, 20–

35% hemicellulose, and 5–30% lignin (10). Because of its recalcitrance, the complex mixture of polymers in

lignocellulose has to undergo a harsh pretreatment processes in order to release the fermentable sugar

monomers. The resulting hydrolysate contains sugars such as hexoses (e.g. glucose) and pentoses (e.g. xylose

and arabinose) (9, 10).

Another promising alternative raw material is the trivalent alcohol glycerol. The growing global demand for

renewable fuels led to strong growth in biodiesel production from vegetable oils mainly in Europe and the United

States (11). Often, biodiesel is obtained from rapeseed oil by transesterification with methanol. During this

process, glycerol is the major by-product with 10% (v/v). Due to the increasing production of biodiesel there is

an excess of glycerol on the marked and valorization is highly desired (12–14). Glycerol is moreover attractive

since its high degree of reduction provides more reducing power per carbon equivalent than other carbon

sources e.g. glucose (15).

The use of these alternative substrates in biotechnological processes, makes highly efficient microbial cell

factories indispensable. As discussed in Chapter 2, several of the traditionally applied cell factories, including

S. cerevisiae, are naturally not able to utilize pentose sugars and glycerol (Figure 5).

15

Figure 5: Substrate spectrum of wild type strains from the commonly used cell factories E. coli, C. glutamicum and

S. cerevisiae. Green: natural substrate. Red: non-natural substrate. Orange: strain dependent substrate usage. Modified

from Buschke, Schäfer, Becker, & Wittmann (2013).

In order to harness these alternative substrates biotechnological engineers have basically two options: The heavy

use of metabolic engineering in order to broaden the substrate capacity of a platform organism or, alternatively,

the selection of an organism, which is naturally able to utilize the desired carbon substrates. Broadening the

substrate spectrum has been done extensively in S. cerevisiae (17, 18), however, this approach has several

drawbacks: Engineering catabolic pathways often result in cofactor imbalances (see Chapter 2). Furthermore,

the heterologous pathway has to be integrated in the cells large regulatory infrastructure and often stress and

starvation-like responses are triggered when an engineered strain grows on non-native substrates (19). The use

of a non-established microorganism with a high natural growth capacity on the alternative carbon source allows

to circumvent these problems. Nevertheless, different problems can arise from the lack of physiological and

genetic knowledge. Fortunately, modern analysis and characterization methods enable to gain these information

in a shorter amount of time than previously (see Chapter 3). In this PhD thesis the yeast Yarrowia lipolytica was

investigated in its ability to utilize the alternative substrates glycerol and the lignocellulosic sugars glucose, xylose

and arabinose. The species and its metabolism is presented in the next chapter.

16

Chapter 2: Yarrowia lipolytica and review of its

carbon metabolism

2.1 Y. lipolytica: A promising host for novel biotechnological applications

Y. lipolytica is a hemiascomycetous yeast which was formally called Candida lipolytica, Endomycopsis lipolytica,

and Saccharomycopsis lipolytica. The genus Yarrowia was identified by David Yarrow in 1972 and reclassified by

Walt and von Arx in 1980 (20, 21). The species name ´lipolytica´ comes from its ability to hydrolyze lipids

efficiently. In order to differentiate Y. lipolytica (and other yeasts) from the well-described species S. cerevisiae

and S. pombe, the term non-conventional yeasts was introduced. Y. lipolytica is dimorphic and can undergo a

true yeast-hyphae transition (22) (Figure 6). Since it differs in several physiological, metabolic and genomic

aspects from S. cerevisiae, it has been used as a model organism, e.g. for the investigation of protein secretion,

hydrophobic substrates utilization, lipid body biogenesis or alternative splicing (23). Y. lipolytica appears in

numerous environments, most of them rich in lipids or fats. For instance, strains have been isolated from food

sources (dairy products and meat), soil, sewage and oil-polluted environments (24–27).

17

Figure 6: Dimorphic character of Y. lipolytica. Depending on environmental conditions this species can grow in the yeast form (upper row) or in the hyphae form (bottom row). The figure shows macroscopic images of Y. lipolytica colonies (A, B), microscopic images of the colony border (C, D) and images of individual cells in liquid culture (E, F).

Y. lipolytica gathered attention as a potential host for biotechnological applications since its early discovery:

Between the 1950s and 1970s Y. lipolytica was used for single-cell protein (SCP) production by British Petroleum

(BP) (27). As mentioned above, it exhibits remarkable lipolytic but also proteolytic activity, due to the secretion

of extracellular enzymes (lipases and proteases). Y. lipolytica is an oleaginous yeast species. Oleaginous

organisms possess a specialized physiology enabling the synthesis and accumulation of high amounts of storage

lipids (28). Furthermore, Y. lipolytica is also a natural producer of organic acids (e.g. citric and isocitric acid, α-

ketoglutaric acid) and sugar alcohols (e.g. mannitol, erythritol) (23, 29). In the last years, the number of

publications related to Y. lipolytica cell factory applications raised continuously (Figure 7). Different strains have

been applied by the research community and inter-strain variations have been observed (30–32). Frequently

applied strains are the French W29, the German H222, the American CBS6124-2, the Polish A-101 and the

Chinese WSH-Z06.

18

Figure 7: Number of PubMed database hits searching for Y. lipolytica.

With increasing availability of genetic tools and sequenced genomes, Y. lipolytica became also tractable for

metabolic engineering approaches. Various studies aimed on broadening Y. lipolytica ´s product spectrum and

substrate range. Limitations of the early genetic tools have been overcome, and today, a comprehensive

synthetic biology toolbox, including CRISPR/CAS genome editing, enables rapid strain development (33). The first

reconstructed Y. lipolytica genome was published in the year 2004 (34) and in the meanwhile several other full

and draft genomes are available (35–38). Genome-scale metabolic models (GEMs) have been generated, which

additionally facilitate the global understanding of Y. lipolytica s metabolism (33).

2.2 Glycerol metabolism, transport and regulation

2.2.1 Yeast growth characteristics on glycerol

The ability to utilize glycerol is characterized by a high intra- and inter-species diversity among yeasts (17). Some

yeasts are not able to use glycerol, others exhibit growth rates similar to those on glucose. Also within a species,

strains can differ drastically in their ability to use this substrate. S. cerevisiae is the most used yeast cell factory

in biotechnological applications, but natural ability to use glycerol is low. A growth assessment of various strains

(natural isolates, common laboratory strains and industrial strains) demonstrated that many S. cerevisiae strains

are not able to grow on glycerol (glycerol- strains), whereas others exhibited moderate growth rates (up to 0.15

h-1) (39). Even though S. cerevisiae possess an inherent potential to grow on glycerol, several non-conventional

19

yeast species exhibit a significantly higher natural capacity to use this substrate. Among others Pachysolen

tannophilus, Yarrowia lipolytica, Pichia pastoris, Pichia anomala and Cyberlindnera jadinii are known species with

a superior growth phenotype on glycerol (40, 41). A direct comparison between S. cerevisiae and several of these

non-conventional yeasts species was performed by Klein et al. (2016). In this study, growth rates of two non-

conventional yeast species (C. jadinii and Y. lipolytica) exceeded 0.4 h-1, whereas for S. cerevisiae growth was

absent or in the range of 0.1 h-1 (Figure 8). Yeast species with a superior glycerol growth phenotype are known

for years, however, the underlying metabolic mechanisms leading to the strong growth are still not fully

elucidated. It has been speculated that the preference for glycerol can be linked to the ecological niche taken by

a certain species (17), as in the case of Y. lipolytica, which can be isolated from lipid rich environments. These

environments exhibit also an high availability of glycerol (42).

Figure 8: Growth comparison of different yeast species on glycerol at two pH levels. Data was taken from Klein et al. 2016.

20

2.2.2 Glycerol catabolic and anabolic routes in fungi

Glycerol metabolism involves all biochemical reactions converting glycerol into central carbon intermediates

(glycerol catabolism) or synthesizing glycerol from precursors (glycerol anabolism). Different glycerol metabolic

pathways exist in yeasts and filamentous fungi, which can be named by their central intermediates (17): glycerol-

3-phosphate (G3P), dihydroxyacetone (DHA) or glyceraldehyde (GA). An overview of the different glycerol

metabolic pathways is shown in Figure 9.

Glycerol catabolism

The glycerol catabolic G3P pathway, also known as the phosphorylative pathway, starts with the phosphorylation

of glycerol to glycerol-3-phosphate by the enzyme glycerol kinase (GK, EC 2.7.1.30) followed by the oxidation to

dihydroxyacetonephosphate (DHAP) by the glycerol-3-phosphate dehydrogenase (mG3PDH, EC 1.1.5.3). The

latter enzyme is bound to the inner mitochondrial membrane and is FAD+-dependent. The catabolic DHA

pathway, also known as the oxidative pathway, starts with the oxidation of glycerol to dihydroxyacetone by an

NAD+-dependent glycerol dehydrogenase (GDH, EC 1.1.1.6) which is followed by a phosphorylation to DHAP. The

latter step is catalyzed by the enzyme dihydroxyacetone kinase (DAK, EC 2.7.1.29). Both pathways (catabolic G3P

and DHA) are leading to DHAP, which is an intermediate of glycolysis and gluconeogenesis. DHAP connects

glycerol catabolism with the central carbon metabolism. An additional metabolic route, the catabolic GA

pathway, has not been described in yeasts so far but has been postulated on the basis of findings in filamentous

fungi (17). Here, glycerol is first oxidized to D-glyceraldehyde (GA) by an NADP+-dependent glycerol

dehydrogenase (GDH, EC 1.1.1.72 / 1.1.1.372). The intermediate GA is proposed to take two potential routes for

entering the central carbon metabolism. It can be either phosphorylated by the enzyme glyceraldehyde kinase

(triokinase, EC 2.7.1.28) or alternatively be oxidized by an aldehyde dehydrogenase (ALDH, EC 1.2.1.3) resulting

in glyceraldehyde-3-phosphate and D-glycerate respectively. Glyceraldehyde-3-phosphate is a glycolytic

intermediate. D-glycerate has to be phosphorylated to 3-phosphoglycerate by the enzyme glycerate kinase (EC

2.7.1.31) before it can enter glycolysis. The pathway was proposed due to measurement of glyceraldehyde kinase

activity in Neurospora crassa mutants able to grow on glycerol (43). Interestingly, homologs for the glycerol

dehydrogenase can be found in S. cerevisiae and Y. lipolytica.

21

Figure 9: Glycerol catabolic and anabolic pathways in fungi. (A) Glycerol-3-phosphate (G3P) pathway, (B) Dihydroxyacetone (DHA) pathway, and (C) Glyceraldehyde (GA) pathway. The GA pathway is so far hypothetical. Glycerol metabolism has been best investigated in S. cerevisiae and corresponding proteins are shown in green. This species uses the G3P pathway but enzymes catalyzing reactions from the other pathways have been identified. Information are taken from Klein et al. (2017).

22

Historically, glycerol catabolic pathways were assigned to a certain yeast species based on key enzyme in vitro

activities. Accordingly yeasts were grouped into containing the G3P pathway, the DHA pathway or both pathways

(44, 45). Later, these biochemical methods were expanded by genetic approaches, in which glycerol pathway

mutants were analyzed, leading to an enhanced understanding. Most of the studies elucidating glycerol

metabolism were conducted with S. cerevisiae. It was demonstrated that S. cerevisiae uses the catabolic G3P

pathway, since the deletions of the genes encoding glycerol kinase (GUT1) or mitochondrial G3P dehydrogenase

(GUT2) completely abolished growth on glycerol (46–48). The role of the catabolic DHA pathway in S. cerevisiae

was ambiguous for a long time and is still debated today (17). Although two isogenes (DAK1 and DAK2) encoding

the enzyme dihydroxyacetone kinase (DAK) could be identified and significant DAK activity measured in vitro (49,

50), no gene or enzymatic activity of the first pathway step (NAD+-dependent glycerol dehydrogenase) could be

detected (49, 51, 52). It has been speculated if the NAPD+-depended dehydrogenases Gcy1p or Ypr1p (GDH EC

1.1.1.72/1.1.1.372) could catalyze the oxidation from glycerol to DHA. However, substrate specificities of these

enzymes are higher for glyceraldehyde (putative GA pathway). Since the reduction of glyceraldehyde to glycerol

is favored, it is unlikely that these enzymes catalyze the step of a potential DHA pathway (49, 51, 53).

Glycerol anabolism

The de novo synthesis of glycerol is crucial, since this molecule fulfills several important cellular functions:

Glycerol acts as the backbone in phospholipids (e.g. phosphatidylcholine), which are important membrane

components. Furthermore, many yeasts use glycerol for osmoregulation and redox balancing. Glycerol

anabolism comprises the backward reactions of the above described catabolic route. These steps are usually

catalyzed by different enzymes.

The anabolic G3P pathway is the main route for glycerol synthesis in S. cerevisiae (54) and is the best studied

route for glycerol synthesis in yeasts. First, DHAP gets reduced to G3P by a cytosolic G3P dehydrogenase

(cG3PDH). This is followed by a dephosphorylation to glycerol catalyzed by a Glycerol-3-phosphatase (GPP, EC

3.1.3.21). G3P dehydrogenases are either NAD+-dependent (EC 1.1.1.8) as in the case of S. cerevisiae and several

other yeasts (55–58) or NADP+-dependent (EC 1.1.1.94) as in the case of Candida versatilis (59). Two isoenzymes

exists in S. cerevisiae for each step encoded by GPD1/GPD2 and GPP1/GPP2 for cG3PDH and GPP respectively.

The mitochondrial and cytosolic G3P dehydrogenases participate together in the so called glycerol-3-phosphate

shuttle, which is important for NAD+ regeneration. As reviewed by Klein et al. (2017) the anabolic DHA pathway

is another route for the synthesis of glycerol in yeasts and fungi. The pathway has been postulated based on the

identification of a NADP+-dependent glycerol dehydrogenase (EC 1.1.1.156) in S. pombe, A. nidulans, A. niger,

23

A. oryzae, and H. jecorina (60–65). This enzyme preferably reduces DHA to glycerol in contrast to glycerol

dehydrogenases from type EC 1.1.1.72 and 1.1.1.372, which prefer the reduction of D-glyceraldehyde to glycerol.

However, the first pathway step requires the dephosphorylation of DHAP to DHA (sugar phosphatase activity: EC

3.1.3.23) which could not be characterized so far. The presence of a functional anabolic DHA pathway was

confirmed in A. nidulans (58, 62). The reverse reactions of the catabolic GA pathway is a third theoretical pathway

for glycerol synthesis (17).

Glycerol metabolism in Y. lipolytica

A great number of studies aimed on the conversion of glycerol or raw glycerol into value-added products by

Y. lipolytica (29), but fewer studies focus on a systematical investigation of the underlying glycerol metabolism.

It is generally accepted that Y. lipolytica uses the catabolic G3P pathway in order to grow on glycerol (66–68).

Blast searches with genes from S. cerevisiae resulted in homologs for glycerol kinase (YlGUT1) and G3P

dehydrogenase (YlGUT2). Dulermo and Nicaud (2011) suggested, that compared to S. cerevisiae Y. lipolytica

possesses a modified and unique glycerol metabolism: only one cytosolic G3P dehydrogenase homolog (YlGPD1)

can be found in Y. lipolytica compared to two isogenes in S. cerevisiae (GPD1/GPD2). Additionally, no glycerol-3-

phosphatase (GPP) homolog could be identified in Y. lipolytica whereas S. cerevisiae again has two isogenes

(GPP1/GPP2). In contrast, numerous homologs to the S. cerevisiae GCY1 and YPR1 genes encoding NADP+-

dependent glycerol dehydrogenases are present in the Y. lipolytica genome, but the function of these

dehydrogenases remain unknown. The authors suggested that Y. lipolytica´s glycerol metabolism is optimized

for the production of G3P, potentially explaining the oleaginous character of this species. A study by Makri, Fakas,

and Aggelis (2010) confirmed the presence of the catabolic G3P pathway on a biochemical level. High activities

of glycerol kinase and G3P dehydrogenase were measured, while no glycerol dehydrogenase activities could be

detected, suggesting the absence of a catabolic DHA pathway.

A few studies exist in which Y. lipolytica mutants impaired in the G3P catabolic pathway were analyzed. An

investigation by Mori et al. (2013) demonstrated that YlGUT1, as well as YlGUT1/YlGUT2 mutants were

strongly growth impaired in media containing glycerol as the only carbon source, however, a slight growth was

still observable. This is in contrast to S. cerevisiae where both GUT1 and GUT2 lead to complete abolishment

of growth. The authors speculated, that the remaining faint but distinct growth could point to an active catabolic

DHA pathway.

24

2.2.3 Glycerol uptake mechanisms

Glycerol has to cross the cell´s plasma membrane before it can get catabolized. Glycerol uptake mechanisms

have been almost exclusively described for S. cerevisiae. These mechanisms were considered to be the main

reason for the observed differences between S. cerevisiae and the superior glycerol utilizing yeasts. A study by

Gancedo et al. (1968) demonstrated that glycerol uptake is 105 times lower in S. cerevisiae than in C. jadinii.

Initially, different glycerol uptake mechanisms were discussed for S. cerevisiae. Initially, glycerol uptake based on

facilitated diffusion by channel proteins has been considered. S. cerevisiae possess the protein Fps1 which is part

of the major intrinsic protein (MIP) family (71). This protein is highly similar to the glycerol facilitator of E.coli

GlpF which is the only uptake system for glycerol in this species (72–74). Initially it was assumed that Fps1p

contributes significantly to glycerol uptake in S. cerevisiae. Later on, it could be shown that Fps1p controls the

efflux of glycerol during osmoregulation rather than being involved in glycerol uptake (75).

Sutherland et al. (1997) predicted that S. cerevisiae uses an Fps1p-independend glycerol/H+ symport system for

the active uptake of glycerol. A study by Ferreira et al. (2005) confirmed this hypothesis: Stl1p a sugar transporter

family member was identified and verified to be responsible for active glycerol transport in S. cerevisiae. Uptake

of glycerol was completely abolished by deleting STL1, preventing mutants to grow on glycerol (77, 78). Glycerol

uptake also via symport (also with other ions e.g. Na+) is also known from other yeast species, and is often

coupled to osmoregulation (79, 80). It should be mentioned, that two membrane proteins, Gup1p and Gup2p

(GUP: Glycerol UPtake), have been considered to be potential glycerol transporters in S. cerevisiae (81). However,

these proteins are nowadays not considered to be glycerol transporters anymore (82).

Glycerol uptake mechanisms have not been systematically investigated in non-S. cerevisiae yeast species yet and

no dedicated work has been conducted in order to systematically investigate glycerol uptake mechanisms in

Y. lipolytica. The increasing amount of non-conventional yeast genomes, allows to conduct homology searches

using S. cerevisiae STL1 and FPS1. Interestingly, several of these non-conventional species contain a higher

number of putative glycerol transporters than S. cerevisiae. For example, D. hansenii possess 8 putative

glycerol/H+ symporters (83). Y. lipolytica possess several putative glycerol transporters: Two homologs of the

S. cerevisiae glycerol facilitator FPS1 are present (YlFPS1 and YlFPS2) and even six homologs of the glycerol/H+

symporter STL1 can be detected by BLAST searches. Two studies demonstrated, that the putative glycerol

facilitators from non-conventional yeasts have different functions than their corresponding homologs in

S. cerevisiae. The FPS1 homolog PtFPS2 from P. tannophilus is one of the most upregulated genes when grown

25

on glycerol compared to glucose (84). Additionally, heterologously expressed PtFPS2 is able to restore growth

defects of an S. cerevisiae STL1 deletion mutant, contrary to the native FPS1 gene. Putative glycerol facilitators

from other yeast species (P. pastoris, C. jadinii and Y. lipolytica) can also significantly improve glycerol uptake in

S. cerevisiae (41, 84). This implies true transporter functions of these FPS1 homologs, rather than controlling

glycerol efflux only as in the case of the S. cerevisiae protein. As reviewed by Klein et al. (2017), the functional

difference is also reflected on a molecular level. The similarity between ScFPS1 and PtFPS2 is mainly restricted

to the six core transmembrane domains and the protein length varies drastically, with 669 amino acid residues

for ScFPS1 compared to 323 residues for PtFPS2. It could be shown that the C- and N-terminus of ScFPS1 are

involved in the closing mechanism (75), which would be dispensable if the channel protein is exclusively

designated for glycerol uptake. Further research has to be done in order to elucidate which transporter functions

as the major glycerol transporter in Y. lipolytica or if an interplay between several transport proteins lead to the

strong growth phenotype.

2.2.4 Glycerol carbon regulation

Microorganisms usually prefer one carbon source, whose presence prevents the utilization of other, alternative

carbon sources. The regulatory mechanisms behind this phenomenon is referred to as carbon catabolite

repression. Sensing the level of the preferred carbon source extra- and intracellularly, repressing genes encoding

enzymes for the degradation of the alternative carbon sources, and initiating the metabolic shift after depletion

of the preferred substrate, are highly complex processes, including many signaling pathways and regulatory

interactions (85).

Glycerol carbon regulatory mechanisms have been extensively investigated in S. cerevisiae but studies are very

rare for other species. S. cerevisiae is known for its ability to convert glucose in a highly efficient manner. Even

under aerobic conditions, S. cerevisiae exhibits primarily alcoholic fermentation, if the glucose concentration

exceeds a certain level. This physiological characteristic is known for a long time and is referred to the Crabtree

effect (86). The main fermentation product is ethanol, but also glycerol is produced in smaller amounts. These

fermentation products are accumulating extracellularly, because the presence of glucose prevents their

utilization. After glucose depletion, the cells switch to a respiratory metabolism and the fermentation products

are re-utilized. The shift between fermentative and respiratory metabolism is characterized by a transition period

26

with a delayed biomass accumulation. This interval is referred to the diauxic shift, and is characterized by the

reorganization of cellular metabolism.

Several studies demonstrated, that the regulation of the carbon catabolite repression mechanism in S. cerevisiae

occur mainly on the level of transcription. Global analysis approaches using microarray based transcriptomics

have been used to get insights into the change of gene expression (87–89) and Klein et al. (2017) reviewed these

studies: As expected, the expression of various genes changed significantly by switching from fermentative

metabolism (glucose utilization, ethanol and glycerol production) to respiratory metabolism (utilization of the

fermentation products). A functional enrichment analysis revealed that many gene-sets related to energy

metabolism and biosynthesis were affected. Highly upregulated genes were related to mitochondrial functions,

energy metabolism, gluconeogenesis, tricarboxylic acid (TCA) cycle, glyoxylate cycle, carbohydrate storage and

stress response. Downregulated genes were related to biosynthesis (transcription by RNA polymerase I and III,

DNA replication and ribosome biogenesis), reflecting the effect of a generally lower growth rate on the

respiratory carbon sources. Transcript levels of the glycerol/H+ symporter (STL1), glycerol kinase (GUT1) and

mitochondrial G3P dehydrogenase (GUT2) strongly increased under respiratory metabolism. Contrary, genes

encoding proteins involved in the anabolic G3P pathway, namely for the cytosolic G3P dehydrogenase (GPD2)

and glycerol-3-phosphatase (GPP1, GPP2) were downregulated. Interestingly, the transcription of the gene

encoding the cytosolic G3P dehydrogenase isoenzyme (GPD1) increased under growth on glycerol and ethanol,

suggesting together with strong GUT2 expression, high activity of the glycerol-3-phosphate shuttle (89). NADP+-

dependent glycerol dehydrogenase (GCY1), whose function remains unclear in S. cerevisiae, was upregulation

under glycerol but not under ethanol.

Also the molecular basis of carbon catabolite repression has been investigated in S. cerevisiae and many

mechanisms could be elucidated (85, 90). However, as reviewed by Klein et al. (2017), knowledge about the

regulation of specific genes involved in glycerol metabolism is still fragmentary. During growth on glucose GUT1

and GUT2 are repressed by the transcriptional regulator Opi1p (91, 92). The transcriptional activators Cat8p and

Adr1p are involved in the derepression of STL1 and GUT1 respectively (91, 93). Derepression of GUT2 involves

the protein kinase Snf1p as well as the Hap2-Hap5 protein complex (91). Other factors crucial for the regulation

of glycerol metabolism in S. cerevisiae (e.g. the balancing of glycerol anabolic and catabolic pathways) are Rsf1p,

Rsf2p (Zms1p) and Hap4p (94–96).

The regulatory network controlling the utilization of glycerol and other carbon sources in Y. lipolytica has not

been systematically analyzed yet. Future research in this area is crucial, since regulatory mechanisms different

27

to those in S. cerevisiae can be expected. Y. lipolytica is a Crabtree-negative yeast exhibiting a strictly respiratory

metabolism and no production of ethanol (97). Furthermore, even though glucose is a well utilized substrate,

several studies showed that glycerol is the preferred carbon source of this species. In single carbon cultivations,

growth rates on glycerol were higher than those on glucose. Furthermore, glycerol is consumed first in glycerol-

glucose mixed cultivations, while the consumption of glucose is suppressed (or delayed) until glycerol has been

depleted (67, 98–101). The glucose suppression by glycerol is shown in Figure 10. Similar effects have also been

reported for other carbon mixtures such as glycerol/acetate (98). The mechanisms restricting the utilization of

other carbon sources in the presence of glycerol are so far unknown, but the observations arise the question, if

this glycerol repression-like effect occurs on the transcriptional level. Findings from other studies support this

hypothesis: It could be shown that the expression of genes involved in the utilization of hydrophobic carbon

sources (n-alkane assimilation) is transcriptionally repressed by glycerol (67, 102, 103). The repressed genes ALK1

and PAT1, encode a cytochrome P450 and an acetoacetyl-CoA thiolase respectively. Until today, reports of

glycerol catabolite repression are rare. One example is the glycerol induced repression of glucose utilization in

the haloarchaeon Haloferax volcanii (104). Indications on glycerol repression mechanisms must be corroborated

by additional experiments. The above mentioned studies demonstrating glycerol repression of n-alkane

assimilation were based on rather old northern blot hybridization methods. Nowadays, more precise and

quantitative methods for the analysis of gene expression are available. These include real-time quantitative PCR

(qPCR), DNA microarrays or RNA-sequencing (RNAseq). A systematic investigation is needed in order to get

insights into the so far totally unknown mechanisms of glycerol repression.

Figure 10: Glycerol-glucose mixed substrate cultivation of Y. lipolytica (98). Glucose utilization is suppressed in the presence of glycerol.

28

2.3 Pentose metabolism

The hydrolysis of lignocellulosic plant material in biorefineries results in fermentable sugar monomers, which

includes beside hexoses also the pentose sugars D-xylose and L-arabinose. In nature, several pentose catabolic

pathways have evolved, which differ to some extend between fungal and bacterial species. Nowadays, most of

the corresponding genes and enzymes have been characterized leading to an almost complete picture of the

pentose metabolism (Bernhard Seiboth 2011). Figure 11 shows the metabolic routes for D-xylose and L-arabinose

assimilation in fungal organisms.

Xylose assimilation in fungal organisms primarily takes place over the oxidoreductase pathway. This pathway

starts with the reduction of D-xylose to D-xylitol catalyzed by the NAD(P)H dependent enzyme xylose reductase

(XYL1, EC 1.1.1.21). D-xylitol is subsequently oxidized to D-xylulose by the NAD(P)+ dependent enzyme xylitol

dehydrogenase (XYL2, EC 1.1.1.9). In the last step D-xylulose is phosphorylated to D-xylulose-5-phosphate by the

enzyme xylulose kinase (XYL3, EC 2.7.1.17). Another route can be found in bacteria, where D-xylose is directly

converted to D-xylulose by the enzyme xylose isomerase (EC 5.3.1.5). This enzyme is not cofactor dependent (18,

19). Xylulose-5-phosphate enters the central carbon metabolism over the non-oxidative branch of the pentose

phosphate pathway (PPP) catalyzed by the enzymes transketolase (TKL, EC 2.2.1.1) and transaldolase (TAL, EC

2.2.1.2).

29

Figure 11: Overview of the pentose catabolic pathways in fungi. D-Xylose (green) and L-arabinose (purple) assimilation

pathways. Abbreviations: XYL1, xylose reductase; XYL2, xylitol dehydrogenase; XYL3, xylulokinase; TKL, transketolase; TAL,

transaldolase; ARD, arabinose reductase; ADH, arabitol dehydrogenase; XLR, xylulose reductase. The figure has been taken

from (105).

In Fungi, L-arabinose is assimilated by another oxidoreductive route which is however interconnected with the

D-xylose degradation pathway. In the first step L-arabinose is reduced to L-arabitol by the NAD(P)H dependent

enzyme L-arabinose reductase (ARD, EC 1.1.1.21). L-arabitol is further oxidized to L-xylulose by the L-arabinitol

dehydrogenase (ADH, EC 1.1.1.12), which is NAD(P)+ dependent. In a second reduction step L-xylulose is

converted into D-xylitol catalyzed by the enzyme L-xylulose reductase (XLR, EC 1.1.1.10). The steps from D-xylitol

to D-xylulose-5-phosphate are catalyzed by the enzymes from the xylose pathway (106).

30

S. cerevisiae is naturally not able to utilize pentose sugars and huge efforts have been undertaken to integrate

the above mentioned pathways into S. cerevisiae cell factories. Xylose utilization has been prioritized since plant

biomass contains less arabinose and the arabinose catabolic pathway includes more enzymatic steps (18). A

model organism for xylose catabolism is the yeast Scheffersomyces stipites which possesses the fungal

oxidoreductive pathway. Expressing corresponding genes in recombinant S. cerevisiae strains enabled xylose

utilization (107, 108). However, the generated strains suffered from co-factor imbalance so that the expression

of the bacterial xylose isomerase became an alternative approach (18, 19).

As in the case of S. cerevisiae, most studies addressing pentose utilization in the non-conventional yeast

Y. lipolytica aimed on the utilization of xylose. There are conflicting reports in the literature if Y. lipolytica is able

to grow on xylose as the sole carbon and energy source. Tsigie et al. (2011) showed, that Y. lipolytica is able to

grow well on xylose. On a mixture of glucose, xylose and arabinose, co-consumption was observed and xylose

consumption was even higher than the consumption of glucose. However, more recent studies reported that

wild type Y. lipolytica strains do either not (110–112) or only use xylose after an adaption phase was carried out

(113). However, it was also reported that several Y. lipolytica strains were able to use xylose when another

carbon sources are present (e.g. glucose). Genome mining indicated that Y. lipolytica contains the

oxidoreductase pathway (XYL1-3) and further biochemical and complementation tests of the candidate enzymes

confirmed the functionality of this metabolic route (112, 113). However, the xylose catabolic pathway in

Y. lipolytica seems to be predominantly cryptic since transcriptional activation of the involved genes is

insufficient. Gene expression analyses led to inconsistent results and seemed to be strain depended. One study

demonstrated an increase of the relative XYL1, XYL2, XYL3, TKL and TAL expression levels when growth on xylose

was compared to glucose (113). In contrast, another study claimed the absence of an inductive effect on xylose

(110).

Only very few studies addressed the utilization of arabinose in Y. lipolytica. One of the very few studies was

conducted by Ryu, Trinh, and Elliot (2018): Genome mining revealed that Y. lipolytica possess putative genes of

the fungal arabinose pathway (ARD, ADH and XLR). Furthermore, a targeted transcriptome approach

demonstrated that several putative genes of the arabinose pathway were upregulated when grown on arabinose

compared to xylose. However, the authors claimed that the enzyme arabitol dehydrogenase is the rate limiting

and prevents efficient arabinose consumption.

31

2.4 Sugar transport in Y. lipolytica

Sugar transport mechanisms have been well described in S. cerevisiae, but are still poorly understood in the non-

conventional yeast Y. lipolytica. In S. cerevisiae hexose uptake is mediated by a single group of homologous

proteins (114). Twenty genes encode proteins related to hexose transport from which 18 are transporters (HXT1-

17 and GAL2) and two are glucose sensors (SNF3 and RGT2). Sugar uptake in S. cerevisiae is highly regulated. The

HXT transporters differ in their substrate affinity and the measurement of the extra- and intracellularly sugar

concentrations triggers expression of the appropriate transporter.

One of the few large scale investigations on hexose uptake in Y. lipolytica has been conducted by Lazar et al.

(2017): Initially, candidate 24 proteins have been identified by homology search using sugar porters of the well

described species S. cerevisiae and K. lactis. Functional analysis is complicated since members of sugar porter

(SP) family comprises hexose transporters (Hxt1-7, Gal2) but also transporters for di- and tri-saccharides (e.g.,

maltose), aliphatic or cyclic polyols (e.g., glycerol or inositol), and several transporters of unknown function (115).

Therefore, the transporters were heterologously expressed in a S. cerevisiae HXT-null mutant, not able to grow

on hexose sugars. Six of the candidate genes functioned as hexose transporters in the heterologous host and

these genes were named Yarrowia Hexose Transporter (YHT1 to YHT6). The YHT genes were further assessed in

deletion and transcriptional studies in Y. lipolytica. It could be shown that Yht1 and Yht4 are likely the main

hexose transporters in Y. lipolytica, and that the other four transporters have “reservoir functions”, with so far

unknown physiological roles.

Furthermore, the authors conducted a phylogenetic analysis including candidate transports from Y. lipolytica and

known transporters from S. cerevisiae and K. lactis. It could be shown that the Y. lipolytica proteins clustered in

six different groups (cluster A to F in Figure 12). Interestingly, in contrast to S. cerevisiae transporters, the bona

fide Y. lipolytica hexose transporters appeared not in one but in three distinct phylogenetic clusters. Cluster A

comprises mainly the S. cerevisiae HXT-type transporters and this group is seems to be not essential in

Y. lipolytica since it only includes Yht3, which is dispensable for the growth on hexoses. Cluster B is

phylogenetically related to hexose sensors in S. cerevisiae and in K. lactis. Yht1 and Yht2 are part of this group.

Experiments suggest that these two proteins are true transporters, despite being close related to sensors and

that Y. lipolytica appears to lack this type of hexose sensor. Lastly, cluster F is related to the K. lactis high-affinity

glucose transporter Hgt1 and includes the essential transporter Yht4 but also Yht5 and Yht6. Interestingly, several

putative Y. lipolytica transporters cluster close to the S. cerevisiae glycerol/H+ symporter Stl1 (Cluster E). It is not

known if these proteins are also involved in glycerol uptake in Y. lipolytica.

32

Figure 12: A phylogenetic analysis of putative Y. lipolytica transport proteins (purple) and known transporters from S. cerevisiae (green) and K. lactis (blue). In contrast to S. cerevisiae HXT1-17 and GAL2, transporters from Y. lipolytica do not cluster in a single group. Y. lipolytica bona fide hexose transporters appear in cluster A (YHT3, similar to HXT-like transporters), cluster B (YHT1, YHT2, similar to S. cerevisiae glucose sensors) and cluster F (YHT4, YHT5, YHT6, similar to K. lactis HGT1). Y. lipolytica transporters showing similarities to the S. cerevisiae glycerol transporter STL1 appear in cluster E. The figure was taken from Lazar et al. (2017).

33

Chapter 3: Methods of cell factory characterization

and analysis As outlined in Chapter 1, the aim of any industrial biotechnological application is an efficient and economical

feasible large scale bioprocess. Optimizing the fermentation step is particularly difficult since biological systems

are involved. Older methods for cell factory development were based on random mutagenesis and screening

approaches, while modern methods additionally integrate a large number of in-depth biological information. For

any rational cell factory research and development project characterization and analysis can be seen as the

groundwork on which other methods such as strain design and construction are built on. Cell factory

characterization is based on quantitative physiology and cell factory analysis comprises the wide field of omics

studies, from which genome sequencing and transcriptomics are discussed below.

3.1 Quantitative physiology

Methods of quantitative physiology are used to assess and quantify the cellular performance of a given

microorganism under defined conditions (116). The physiological data enables to estimate key performance

indicators of the process, such as the microbial growth rate, rates of substrate consumption, rates of product

generation or yield coefficients. The gained physiological parameters have wide applications and can be used,

for instance, to compare different candidate strains in screening approaches, to design and evaluate

bioprocesses or to guide up-scaling attempts. The overall cellular performance is the result of a complex interplay

between physical and chemical conditions on one side, and the biological system on the other side. Process

relevant physical, chemical and biological parameters are among others the temperature, ambient pH, oxygen

level, biomass concentration and substrate concentrations, which all must be measured in a precise and robust

way (117, 118). The applied media composition has a tremendous effect on the cellular performance and

methods of quantitative physiology are used in media optimization procedures. Traditionally, complex media

was widely used in industrial settings due to its low price and availability, however, defined minimal media is

preferred in research and development. Here, the exact stoichiometry of each media constituent is known,

increasing the experimental reproducibility and allows a precise quantification of the process, since each element

can be balanced (119).

In order to evaluate the cellular performance of a microorganism cultivation experiments have to be conducted.

These cultivations can take place in various different vessels and volumes. A general problem in the quantitative

34

determination of physiology is throughput versus the level of gained information. Techniques for cultivation

range from micro-titer plates (MTP), over shake flasks, to bioreactors of different complexity. Microtiter plates

allow highly parallelized cultivations, but are limited in monitoring and controlling various important cultivation

parameters. In contrast, a bioreactor set-up is very labor intensive but is equipped with various sensors and allow

a significantly higher degree of control (120). The applied cultivation technique is dependent on the research

question addressed: For example, modern strain engineering methods lead to a significantly increased number

of created strains which need to be characterized (strain screening). High throughput characterization

approaches help to select the best performing strains in a short amount of time (121). On the other side, some

experimental techniques (e.g. omic-related analysis) require tailor-made biomass, which is only obtained from

highly controlled cultivations. Reproducibility, low technical and biological variability of the replicates are needed

for the subsequent statistical analysis (116).

Three different modes of cultivation are commonly applied in laboratory settings: batch, fed-batch and

continuous cultivations. Batch cultivations are used to gain biomass for further experiments (overnight batch

cultures) or as a simple way to receive process information. All nutrients are available in sufficient amounts from

the beginning of an experiment, allowing the cells to grow at an unlimited rate (μmax). This exponential phase

lasts until nutrient depletion or the accumulation of inhibiting metabolites. From a research prospective,

however, the batch culture is a poor experimental tool (122). Biomass, substrate and product concentrations

change over time and constant conditions are limited to short time intervals. The observed phenotype is,

therefore, always influenced by earlier conditions. A modified version of the batch is the so-called fed-batch

culture. Here, the addition of a feed solution allows to work with growth inhibiting substrates (e.g. methanol),

to prevent overflow metabolism or to reach high cell densities (123). As the batch culture, also fed-batch culture

is affected by changing conditions over time. In contrast, the continuous cultivation in chemostats is a uniquely

suited instrument for physiological studies. In chemostat settings, fresh media is added to the reactor and broth

is removed from the reactor with a constant rate. A steady state can be reached in which the physicochemical

conditions (e.g. concentrations) and all rates of production and consumption stay constant (124). Another

important advantage is that the experimenter can control the dilution rate and by that the microorganism’s

growth rate (122). One nutrient is usually limiting in chemostat cultivations, meaning that higher concentrations

of this nutrient result in a higher biomass concentration. All other nutrients are present in excess. Chemostat

cultivations are suited to independently manipulated and examine single process variables. Furthermore, the

continuous culture is uniquely suited for the quantification of maintenance energy requirements. One limitation

35

of the chemostats are the occurring mutation and selection events under steady state conditions. However, this

becomes only significant when the experiments lasts longer than 3-6 days (122).

3.2 Genomics

3.2.1 High-throughput genome sequencing, assembly and annotation

The aim of genome sequencing is the estimation of the complete nucleotide sequence of an organism of interest.

The development of genome sequencing technologies was enabled by a close interaction between various

disciplines such as biology, chemistry, engineering and computer science (125). Early sequencing projects were

limited to model organisms and required high amounts of resources as well as the participation of large consortia

(126). The first complete reconstructed genome, of the bacterium Haemophilus influenza, was published in the

year 1995 and consist of 1,830,140 base pairs of DNA (127). The first eukaryotic genome (S. cerevisiae) was

published one year later and consisted already of 12,000,000 base pairs (128). Through tremendous progress in

both, sequencing technology and data analyses solutions, it became possible to create de novo draft genome

sequences even in individual research groups (126). Nowadays, the genomes of tens of thousands of bacterial

and viral species, thousands of individual humans and hundreds to thousands of genomes from other organisms

have been sequenced (125). The ease to gain genomic information changed the way genomic research is

conducted. Advances in genome sequencing and genomics also have a huge impact on the development and

analysis of cell factories. For example reverse engineering, the introduction of specific mutations identified from

evolutionary engineering experiments, connects classical strain engineering with modern rationale approaches.

Genome sequencing enables further various other applications such as different omics methods or genome-scale

metabolic models. In this section, a typical workflow of a genome sequencing project is presented (Figure 13 A).

Sequencing projects require time, sufficient financial and computational resources and also careful planning (e.g.

coverage considerations) (126). They include various laboratory (“wet-lab”) and computational (“dry-lab”) steps.

The first step in genome sequencing is the isolation of high-quality genomic material. In the case of microbial cell

factories, the strain of interest must be cultured, harvested and undergo genomic DNA extraction protocols.

Nowadays, most genome projects use high-throughput sequencing (HTS), and Illumina provides the most

popular platforms. Roughly sketched, sequencing on Illumina systems include the following steps: sample

preparation (DNA fragmentation, adapter ligation and PCR enrichment), clonal amplification (cluster generation

on a chip), massively parallel sequencing (base incorporation, washing, imaging and cleavage) and data analysis

(cluster identification, base calling and base quality score estimation) (129). Compared to older sequencing

36

methods, HTS approaches have massive upscaling possibilities, resulting in significantly decreased sequencing

times.

After finishing the sequencing process, the data is received in the form of raw reads. The amount of generated

data is huge (hundreds of gigabytes), and downstream analysis is performance intensive. Therefore, it is

recommended to use high-performance computers for the following steps. The assembly pipeline starts with a

read quality control procedure. The raw sequencing reads are encoded in the FASTQ file format, which contains

besides the nucleotide sequence also a quality score at each nucleotide position. Specific software tools like

FastQC (130) generate various summary statistics. These include among others, base quality distributions, GC

content, adaptor contamination and duplicated reads. Usually, the read quality drops towards the end, and

quality trimming is recommended prior to the assembly (Figure 13 B). Additionally, remaining adaptor sequences

are often present in the reads and proper adapter identification and subsequent adapter trimming should be

carried out (131). Sequencing a genome by HTS would be impossible without specialized computational tools,

named genome sequence assemblers. A genome assembler use reads as the input and generates contiguous

sequences named contigs. Assemblers have been developed and optimized over the last 35 years, and different

theoretical approaches have been used (125): Simple so-called greedy approaches iteratively join reads in

decreasing order of their overlaps. Graph-based approaches model the sequence data as graphs and currently,

the most common approach for assembling short read data is based on De Bruijn graphs (126). Here, each read

is broken into overlapping k-mers (substrings of the read), which are added as vertices to the graph. Adjacent k-

mers are linked by edges, and the assembly problem can be formulated as an Eulerian path problem (Figure 13

C). Generating contigs from sequencing reads is called a de novo assembly. Since various assembly tools are

available with different assembly paradigms, it is difficult to predict which tool gives the best output. An assembly

process, therefore, must be treated as an iterative procedure trying out different programs and parameters.

37

Figure 13: Aspects of draft genome sequencing. (A) A typical genome sequencing workflow: Shotgun high-throughput sequencing generates single-end, paired end or mate-pair reads. Genome assemblers generate contigs from the reads, which can be further connected into scaffolds. Genome annotation is the last step in building draft-genomes. (B) The sequence base quality of a read usually drops in the end, making read quality control and trimming necessary. (C) Many modern genome assemblers are based on De Bruijn graphs: Overlapping reads are generated and broken down into k-mers. The k-mers form the edges of the graph and the assembly problem is formulated as an Eulerian path problem. Figure A and C are taken from Ekblom & Wolf (2014), figure B is taken from Andrews (2010).

The contigs can be further assembled into scaffolds (supercontigs) based on read pair information (125, 126).

Gaps between contigs are then usually filled with Ns (variable bases). Since the number of sequenced genomes

is strongly increasing, the scaffolding process can be facilitated if a closely related reference genome (e.g. another

strain of the same species) is available. In this so-called reference assisted genome assembly approach, the

reference genome provides information about contig orientation and relative position in the genome. Usually,

contigs are first generated by de novo read assembly, which are then get aligned to the reference genome.

Genome annotation is the final step in the generation of a draft genome. With the upcoming of next-generation

sequencing technologies automated annotation procedures became necessary. However, genome annotation

still requires manual steps (132). Genome annotation includes the inference of gene structures and the

38

estimation of orthology/paralogy relationships between these new genes and genes of other species (132). Often

a combination of different annotation approaches is used (38): If an annotated reference genome exists, it is

possible to map the existing annotations to the new genome by alignment. For some phyla, specialized

annotation services are available like the YGAP pipeline for yeasts, which is based on homology and conserved

synteny information (132). Furthermore, RNA-seq data can be used to support the identification of coding

regions (CDS) (133). The finished assembled, and annotated genome provides a valuable foundation to address

various research question. It is important to treat the obtained genome sequence as a working hypothesis. Due

to genetic variations between individuals, sequencing errors and assembly errors the ideal of a complete

reconstructed genome of a given species is unattainable in practice (126).

3.2.2 Sequencing-based transcriptomics (RNA-seq)

The transcriptome comprises the sum of all RNA transcripts in a cell and studying the transcriptome is an essential

part of functional genomics. In contrast to the genome which is mainly static, the transcriptome is dynamic and

changes with different developmental states or environmental conditions. Therefore, transcriptomics focuses on

the active part of the genome. RNA-seq has become the method of choice for genome-wide gene expression

studies. It is based on Next-Generation Sequencing and has several advantaged over older methods such as

hybridization-based microarrays technologies (134). Microarrays allow only the investigation of transcripts for

which probes have been designed, while RNAseq analyzes the whole transcriptome. This enables the

examination of additional genomic features such as unknown transcripts, nontranslated regions or alternative

splicing events (134). Furthermore, RNA-seq has a better dynamic range, a higher resolution, and a generally

lower technical variability (135). Transcriptome studies are commonly used in cell factory design and analysis

applications (136): The technology has been used for instance to identify metabolic engineering targets (e.g. a

lowly expressed gene). Also, the effect of a classical strain improvement attempts (e.g. chemical mutagenesis

and screening) can be investigated in the transcriptional level. Usually, the transcriptomes from strains differing

in a specific trait (e.g. wild-type versus producer strain) are compared. A general limitation of transcriptome

studies is the fact, that phenotypic responses are not always linked directly to the transcriptional pattern.

In the following, a typical RNA-seq based workflow for differential expression analysis is outlined. Many steps

are similar to those of genome sequencing (DNA-seq) since both applications are based on high-throughput

sequencing. In an RNA-seq attempt, the formulation of a precise biological question has an important role which

influences the experimental design directly. It is crucial to generating data which allows answering the research

39

question. Among others, considerations concern the RNA molecule of interest (mRNA or others), sequencing

depth, library type or the number of replicates (137). After the raw reads have been generated, quality control

and read trimming steps is necessary which are similar to the above described DNA-seq procedures. Prior to the

core transcriptome analysis, the sequencing reads must be processed and quantified in order to obtain an

expression value for each transcript. This procedure includes two steps which are usually carried out on high-

performance computer systems: First, the reads must be mapped to an annotated reference genome (or

transcriptome) by alignment (137). If no reference is available, the reads can also first be de novo assembled into

contigs and subsequently mapped to this new reference. Mapping is conducted with special read alignment

software, and the output is usually stored in the Sequence Alignment Map or Binary Alignment Map (SAM/BAM)

format (138). Second, the mapping results must be quantified in terms of the amount of reads overlapping a

genomic feature (read coverage) (139). The result of this read summarization is a count matrix with integer

values. In this matrix of integer values, rows represent genomic features genes (e.g. genes) and columns

represent the samples (140). The output of RNA-seq and microarray experiments differs. Microarrays yield in

intensities, which are continuous numerical values. The nature of the output has implications in the downstream

statistical analysis (139).

The obtained count matrix marks the starting point of the core statistical analysis of differential gene expression.

This analysis is usually less computationally intensive and can be conducted on personal computers, equipped

with statistic software like the R programming language. A typical analysis comprises data import and packaging,

data pre-processing, exploratory data analysis, differential expression testing and optionally gene set testing

(141). Various software packages are available for differential gene expression analysis, including popular tools

such as DESeq (142), DESeq2 (143), edgeR (144) or limma (145), which differ in the used statistical models and

assumptions. In this thesis, the analysis was conducted with edgeR, limma and the piano package (Chapter 6).

Data pre-processing includes a step for the removal of genes which are generally low expressed in all conditions.

These genes provide no meaningful information and can complicate the further analysis. Different visualization

methods such as principal components analysis (PCA) or multidimensional scaling (MDS) plots are used in the

exploratory data analysis procedure in order to show similarities and dissimilarities between the samples. This

provides valuable information to what extent differential expression can be expected and if outliers are present.

Several none-biological factors can affect the expression levels of certain samples, and therefore, normalization

is required (141). The expression values of the samples should have a similar range and distribution. In edgeR,

scaling factors for the library size are calculated by the method of trimmed mean of M-values (TMM) (146). After

40

estimating the scaling factors, the first step in the differential expression analysis is the creation of a matrix in

which all relevant experimental information is collected (design matrix). Linear modelling in limma was initially

developed to analyze continuous numerical microarray data, but can also be applied to RNA-seq integer counts,

if the voom function is applied. Voom is an acronym for mean-variance modelling at the observational level. It

was shown that the variance for count values is not independent of the mean, meaning that raw counts exhibit

an increasing mean-variance trend, while log-transformed counts show a decreasing trend (139). This hampers

normal-based statistical methods. Therefore, the voom function converts the counts first into log2-counts per

million (log-CPM), then estimates the mean-variance relationship and finally computes appropriate observation-

level weights. These weights are used in the subsequent linear modelling process to remove heteroscedasticity.

Limma finally applies empirical Bayes moderation to borrow strength across genes, which lead to more precise

estimates of gene-wise variability. The result of a differential gene expression analysis is a table containing gene-

level statistics. For each gene, several statistical matrices are provided, such as p-value, adjusted p-value, log

fold-change or t-statistic. The results can be used to discuss the initially asked biological question. Gene-level

statistics alone, however, does not necessarily facilitate biological interpretations, since a large amount of data

must be analyzed manually and no information about the functional connectedness of differentially expressed

genes is provided. Therefore, specific statistical hypothesis tests have been developed, which combine gene-

level statistics with existing biological knowledge (147). These methods referred to as gene set analysis (GSA),

gene set enrichment analysis or gene set testing, map the gene-level statistics to known biological functions or

processes. An analysis of the resulting significant gene-sets drastically facilitates interpretation of the results.

41

References of Part I

1. El-Mansi EMT, Bryce CFA, Dahhou B, Sanchez S, Demain AL, Allman AR. 2012. Fermentation microbiology and biotechnology. Third edition. Taylor & Francis/CRC Press.

2. Becker J, Wittmann C. 2015. Advanced Biotechnology: Metabolically Engineered Cells for the Bio-Based Production of Chemicals and Fuels, Materials, and Health-Care Products. Angew Chemie Int Ed 54:3328–3350.

3. Hong K-K, Nielsen J. 2012. Metabolic engineering of Saccharomyces cerevisiae: a key cell factory platform for future biorefineries. Cell Mol Life Sci 69:2671–2690.

4. Singh A, Singh B, Ward O. 2012. Potential applications of bioprocess technology in petroleum industry. Biodegradation 23:865–880.

5. Sutcliffe S, Cannon W, van Berkel M, Mosmuller W. 2016. Commercial Success in a Low Oil Price Environment. Ind Biotechnol 12:327–333.

6. EUROPEAN COMMISSION. 2012. Innovating for Sustainable Growth: A Bioeconomy for Europe. Commun FROM Comm TO Eur Parliam Counc Eur Econ Soc Comm Comm Reg.

7. Campbell K, Xia J, Nielsen J. 2017. The Impact of Systems Biology on Bioprocessing. Trends Biotechnol 35:1156–1168.

8. Demeke MM, Dietz H, Li Y, Foulquié-Moreno MR, Mutturi S, Deprez S, Den Abt T, Bonini BM, Liden G, Dumortier F, Verplaetse A, Boles E, Thevelein JM. 2013. Development of a D-xylose fermenting and inhibitor tolerant industrial Saccharomyces cerevisiae strain with high performance in lignocellulose hydrolysates using metabolic and evolutionary engineering. Biotechnol Biofuels 6:89.

9. Weber C, Farwick A, Benisch F, Brat D, Dietz H, Subtil T, Boles E. 2010. Trends and challenges in the microbial production of lignocellulosic bioalcohol fuels. Appl Microbiol Biotechnol 87:1303–1315.

10. Jäger G, Büchs J. 2012. Biocatalytic conversion of lignocellulose to platform chemicals. Biotechnol J 7:1122–36.

11. Valerio O, Horvath T, Pond C, Manjusri Misra, Mohanty A. 2015. Improved utilization of crude glycerol from biodiesel industries: Synthesis and characterization of sustainable biobased polyesters. Ind Crops Prod 78:141–147.

12. da Silva GP, Mack M, Contiero J. 2009. Glycerol: a promising and abundant carbon source for industrial microbiology. Biotechnol Adv 27:30–9.

13. Johnson DT, Taconi KA. 2007. The glycerin glut: Options for the value-added conversion of crude glycerol resulting from biodiesel production. Environ Prog 26:338–348.

14. Pagliaro M, Ciriminna R, Kimura H, Rossi M, Della Pina C. 2007. From Glycerol to Value-Added Products. Angew Chemie Int Ed 46:4434–4440.

15. Klein M, Carrillo M, Xiberras J, Islam Z, Swinnen S, Nevoigt E. 2016. Towards the exploitation of glycerol’s

42

high reducing power in Saccharomyces cerevisiae -based bioprocesses. Metab Eng 38:464–472.

16. Buschke N, Schäfer R, Becker J, Wittmann C. 2013. Metabolic engineering of industrial platform microorganisms for biorefinery applications – Optimization of substrate spectrum and process robustness by rational and evolutive strategies. Bioresour Technol 135:544–554.

17. Klein M, Swinnen S, Thevelein JM, Nevoigt E. 2017. Glycerol metabolism and transport in yeast and fungi: established knowledge and ambiguities. Environ Microbiol 19:878–893.

18. Hahn-Hägerdal B, Karhumaa K, Jeppsson M, Gorwa-Grauslund MF. 2007. Metabolic Engineering for Pentose Utilization in Saccharomyces cerevisiae, p. 147–177. In Biofuels. Springer Berlin Heidelberg, Berlin, Heidelberg.

19. Gopinarayanan VE, Nair NU. 2018. Pentose metabolism in Saccharomyces cerevisiae : The need to engineer global regulatory systems. Biotechnol J 1800364.

20. Yarrow D. 1972. Four new combinations in yeasts. Antonie Van Leeuwenhoek 38:357–60.

21. van der Walt JP, von Arx JA. 1980. The yeast genus Yarrowia gen. nov. Antonie Van Leeuwenhoek 46:517–521.

22. Domínguez A, Fermiñán E, Gaillardin C. 2000. Yarrowia lipolytica: An Organism Amenable to Genetic Manipulation as a Model for Analyzing Dimorphism in FungiContrib Microbiol. Basel, Karger.

23. Nicaud J-M. 2012. Yarrowia lipolytica. Yeast 29:409–18.

24. Beopoulos A, Desfougeres T, Sabirova J, Zinjarde S, Neuvéglise C, Nicaud J-M. 2010. The Hydrocarbon-Degrading Oleaginous Yeast Yarrowia lipolytica, p. 2111–2121. In Handbook of Hydrocarbon and Lipid Microbiology. Springer Berlin Heidelberg, Berlin, Heidelberg.

25. Barth G, Gaillardin C. 1997. Physiology and genetics of the dimorphic fungus Yarrowia lipolytica. FEMS Microbiol Rev 19:219–37.

26. Fickers P, Benetti P, Wache Y, Marty A, Mauersberger S, Smit M, Nicaud J. 2005. Hydrophobic substrate utilisation by the yeast , and its potential applications. FEMS Yeast Res 5:527–543.

27. Groenewald M, Boekhout T, Neuvéglise C, Gaillardin C, van Dijck PWM, Wyss M. 2014. Yarrowia lipolytica: safety assessment of an oleaginous yeast with a great industrial potential. Crit Rev Microbiol 40:187–206.

28. Adrio JL. 2017. Oleaginous yeasts: Promising platforms for the production of oleochemicals and biofuels. Biotechnol Bioeng 114:1915–1920.

29. Rywińska A, Juszczyk P, Wojtatowicz M, Robak M, Lazar Z, Tomaszewska L, Rymowicz W. 2013. Glycerol as a promising substrate for Yarrowia lipolytica biotechnological applications. Biomass Bioenergy 48:148–166.

30. Naumova E, Naumov G, Fournier P, Nguyen H-V, Gaillardin C. 1993. Chromosomal polymorphism of the yeast Yarrowia lipolytica and related species: electrophoretic karyotyping and hybridization with cloned genes. Curr Genet 23:450–454.

31. Lazar Z, Dulermo T, Neuvéglise C, Crutz-Le Coq A-M, Nicaud J-M. 2014. Hexokinase—A limiting factor in lipid production from fructose in Yarrowia lipolytica. Metab Eng 26:89–99.

43

32. Andre A, Chatzifragkou A, Diamantopoulou P, Sarris D, Philippoussis A, Galiotou-Panayotou M, Komaitis M, Papanikolaou S. 2009. Biotechnological conversions of bio-diesel-derived crude glycerol by Yarrowia lipolytica strains. Eng Life Sci 9:468–478.

33. Markham KA, Alper HS. 2018. Synthetic Biology Expands the Industrial Potential of Yarrowia lipolytica. Trends Biotechnol 36:1085–1095.

34. Dujon B, Sherman D, Fischer G, Durrens P, Casaregola S, Lafontaine I, De Montigny J, Marck C, Neuvéglise C, Talla E, Goffard N, Frangeul L, Aigle M, Anthouard V, Babour A, Barbe V, Barnay S, Blanchin S, Beckerich J-M, Beyne E, Bleykasten C, Boisramé A, Boyer J, Cattolico L, Confanioleri F, De Daruvar A, Despons L, Fabre E, Fairhead C, Ferry-Dumazet H, Groppi A, Hantraye F, Hennequin C, Jauniaux N, Joyet P, Kachouri R, Kerrest A, Koszul R, Lemaire M, Lesur I, Ma L, Muller H, Nicaud J-M, Nikolski M, Oztas S, Ozier-Kalogeropoulos O, Pellenz S, Potier S, Richard G-F, Straub M-L, Suleau A, Swennen D, Tekaia F, Wésolowski-Louvel M, Westhof E, Wirth B, Zeniou-Meyer M, Zivanovic I, Bolotin-Fukuhara M, Thierry A, Bouchier C, Caudron B, Scarpelli C, Gaillardin C, Weissenbach J, Wincker P, Souciet J-L. 2004. Genome evolution in yeasts. Nature 430:35–44.

35. Liu L, Alper HS. 2014. Draft Genome Sequence of the Oleaginous Yeast Yarrowia lipolytica PO1f, a Commonly Used Metabolic Engineering Host. Genome Announc 2.

36. Devillers H, Brunel F, Połomska X, Sarilar V, Lazar Z, Robak M, Neuvéglise C. 2016. Draft Genome Sequence of Yarrowia lipolytica Strain A-101 Isolated from Polluted Soil in Poland. Genome Announc 4.

37. Pomraning KR, Baker SE. 2015. Draft Genome Sequence of the Dimorphic Yeast Yarrowia lipolytica Strain W29. Genome Announc 3.

38. Magnan C, Yu J, Chang I, Jahn E, Kanomata Y, Wu J, Zeller M, Oakes M, Baldi P, Sandmeyer S. 2016. Sequence Assembly of Yarrowia lipolytica Strain W29/CLIB89 Shows Transposable Element Diversity. PLoS One 11:e0162363.

39. Swinnen S, Klein M, Carrillo M, McInnes J, Nguyen HTT, Nevoigt E. 2013. Re-evaluation of glycerol utilization in Saccharomyces cerevisiae: characterization of an isolate that grows on glycerol without supporting supplements. Biotechnol Biofuels 6:157.

40. Silva-Graça M, Lages F, Lucas C. 1999. Active glycerol uptake is a mechanism underlying halotolerance in yeasts: a study of 42 species. Microbiology 145:2577–2585.

41. Klein M, Islam Z-U, Knudsen PB, Carrillo M, Swinnen S, Workman M, Nevoigt E. 2016. The expression of glycerol facilitators from various yeast species improves growth on glycerol of Saccharomyces cerevisiae. Metab Eng Commun 3:252–257.

42. Barth G, Gaillardin C. 1997. Physiology and genetics of the dimorphic fungus Yarrowia lipolytica. FEMS Microbiol Rev 19:219–237.

43. Tom GD, Viswanath-Reddy M, Howe HB. 1978. Effect of carbon source on enzymes involved in glycerol metabolism in Neurospora crassa. Arch Microbiol 117:259–63.

44. Babel W, Hofmann KH. 1982. The relation between the assimilation of methanol and glycerol in yeasts. Arch Microbiol 132:179–184.

45. TANI Y, YAMADA K. 1987. Glycerol metabolism in methylotrophic yeasts. Agric Biol Chem 51:1927–1933.

44

46. Sprague GF, Cronan JE. 1977. Isolation and characterization of Saccharomyces cerevisiae mutants defective in glycerol catabolism. J Bacteriol 129:1335–42.

47. Pavlik P, Simon M, Schuster T, Ruis H. 1993. The glycerol kinase (GUT1) gene of Saccharomyces cerevisiae: cloning and characterization. Curr Genet 24:21–5.

48. Swinnen S, Klein M, Carrillo M, McInnes J, Nguyen HT, Nevoigt E. 2013. Re-evaluation of glycerol utilization in Saccharomyces cerevisiae: characterization of an isolate that grows on glycerol without supporting supplements. Biotechnol Biofuels 6:157.

49. Norbeck J, Blomberg A. 1997. Metabolic and regulatory changes associated with growth of Saccharomyces cerevisiae in 1.4 M NaCl. Evidence for osmotic induction of glycerol dissimilation via the dihydroxyacetone pathway. J Biol Chem 272:5544–54.

50. Molin M, Norbeck J, Blomberg A. 2003. Dihydroxyacetone kinases in Saccharomyces cerevisiae are involved in detoxification of dihydroxyacetone. J Biol Chem 278:1415–23.

51. Ford G, Ellis EM. 2002. Characterization of Ypr1p fromSaccharomyces cerevisiae as a 2-methylbutyraldehyde reductase. Yeast 19:1087–1096.

52. Nguyen HTT, Nevoigt E. 2009. Engineering of Saccharomyces cerevisiae for the production of dihydroxyacetone (DHA) from sugars: A proof of concept. Metab Eng 11:335–346.

53. Chang Q, Griest TA, Harter TM, Mark Petrash J. 2007. Functional studies of aldo-keto reductases in Saccharomyces cerevisiae. Biochim Biophys Acta - Mol Cell Res 1773:321–329.

54. Nevoigt E, Stahl U. 1997. Osmoregulation and glycerol metabolism in the yeast Saccharomyces cerevisiae. FEMS Microbiol Rev 21:231–41.

55. Lee D-H, Kim M-D, Ryu Y-W, Seo J-H. 2008. Cloning and characterization of CmGPD1, the Candida magnoliae homologue of glycerol-3-phosphate dehydrogenase. FEMS Yeast Res 8:1324–33.

56. Chen X, Fang H, Rao Z, Shen W, Zhuge B, Wang Z, Zhuge J. 2008. Cloning and characterization of a NAD + -dependent glycerol-3-phosphate dehydrogenase gene from Candida glycerinogenes , an industrial glycerol producer. FEMS Yeast Res 8:725–734.

57. Thomé PE. 2004. Isolation of aGPD gene fromDebaryomyces hansenii encoding a glycerol 3-phosphate dehydrogenase(NAD+). Yeast 21:119–126.

58. Fillinger S, Ruijter G, Tamás MJ, Visser J, Thevelein JM, d’Enfert C. 2001. Molecular and physiological characterization of the NAD-dependent glycerol 3-phosphate dehydrogenase in the filamentous fungus Aspergillus nidulans. Mol Microbiol 39:145–57.

59. Watanabe Y, Nagayama K, Tamai Y. 2008. Expression of glycerol 3-phosphate dehydrogenase gene (CvGPD1) in salt-tolerant yeastCandida versatilis is stimulated by high concentrations of NaCl. Yeast 25:107–116.

60. Marshall JH, Kong Y-C, Sloan J, May JW. 1989. Purification and Properties of Glycerol: NADP+ 2-Oxidoreductase from Schizosaccharomyces pombe. Microbiology 135:697–701.

61. Redkar RJ, Locy RD, Singh NK. 1995. Biosynthetic Pathways of Glycerol Accumulation under Salt Stress in Aspergillus nidulans. Exp Mycol 19:241–246.

45

62. de Vries RP, Flitter SJ, van de Vondervoort PJI, Chaveroche M-K, Fontaine T, Fillinger S, Ruijter GJG, d’Enfert C, Visser J. 2003. Glycerol dehydrogenase, encoded by gldB is essential for osmotolerance in Aspergillus nidulans. Mol Microbiol 49:131–41.

63. Schuurink R, Busink R, Hondmann DHA, Wltteveen CFB, Vlsser J. 1990. Purification and properties of NADP+-dependent glycerol dehydrogenases from Aspergillus nidulans and A. niger. J Gen Microbiol 136:1043–1050.

64. Ruijter GJG, Visser J, Rinzema A. 2004. Polyol accumulation by Aspergillus oryzae at low water activity in solid-state fermentation. Microbiology 150:1095–1101.

65. Liepins J, Kuorelahti S, Penttila M, Richard P. 2006. Enzymes for the NADPH-dependent reduction of dihydroxyacetone and D-glyceraldehyde and L-glyceraldehyde in the mould Hypocrea jecorina. FEBS J 273:4229–4235.

66. Dulermo T, Nicaud J-M. 2011. Involvement of the G3P shuttle and β-oxidation pathway in the control of TAG synthesis and lipid accumulation in Yarrowia lipolytica. Metab Eng 13:482–91.

67. Mori K, Iwama R, Kobayashi S, Horiuchi H, Fukuda R, Ohta A. 2013. Transcriptional repression by glycerol of genes involved in the assimilation of n-alkanes and fatty acids in yeast Yarrowia lipolytica. FEMS Yeast Res 13:233–40.

68. Beopoulos A, Mrozova Z, Thevenieau F, Le Dall M-T, Hapala I, Papanikolaou S, Chardot T, Nicaud J-M. 2008. Control of lipid accumulation in the yeast Yarrowia lipolytica. Appl Environ Microbiol 74:7779–89.

69. Makri A, Fakas S, Aggelis G. 2010. Metabolic activities of biotechnological interest in Yarrowia lipolytica grown on glycerol in repeated batch cultures. Bioresour Technol 101:2351–2358.

70. Gancedo C, Gancedo JM, Sols A. 1968. Glycerol metabolism in yeasts. Pathways of utilization and production. Eur J Biochem 5:165–72.

71. Luyten K, Albertyn J, Skibbe WF, Prior BA, Ramos J, Thevelein JM, Hohmann S. 1995. Fps1, a yeast member of the MIP family of channel proteins, is a facilitator for glycerol uptake and efflux and is inactive under osmotic stress. EMBO J 14:1360–71.

72. Heller KB, Lin EC, Wilson TH. 1980. Substrate specificity and transport properties of the glycerol facilitator of Escherichia coli. J Bacteriol 144:274–8.

73. Sweet G, Gandor C, Voegele R, Wittekindt N, Beuerle J, Truniger V, Lin EC, Boos W. 1990. Glycerol facilitator of Escherichia coli: cloning of glpF and identification of the glpF product. J Bacteriol 172:424–30.

74. Maurel C, Reizer J, Schroeder JI, Chrispeels MJ, Saier MH. 1994. Functional characterization of the Escherichia coli glycerol facilitator, GlpF, in Xenopus oocytes. J Biol Chem 269:11869–72.

75. Tamás MJ, Luyten K, Sutherland FC, Hernandez A, Albertyn J, Valadi H, Li H, Prior BA, Kilian SG, Ramos J, Gustafsson L, Thevelein JM, Hohmann S. 1999. Fps1p controls the accumulation and release of the compatible solute glycerol in yeast osmoregulation. Mol Microbiol 31:1087–104.

76. Sutherland FC, Lages F, Lucas C, Luyten K, Albertyn J, Hohmann S, Prior BA, Kilian SG. 1997. Characteristics of Fps1-dependent and -independent glycerol transport in Saccharomyces cerevisiae. J Bacteriol

46

179:7790–5.

77. Ferreira C, van Voorst F, Martins A, Neves L, Oliveira R, Kielland-Brandt MC, Lucas C, Brandt A. 2005. A member of the sugar transporter family, Stl1p is the glycerol/H+ symporter in Saccharomyces cerevisiae. Mol Biol Cell 16:2068–76.

78. Swinnen S, Klein M, Carrillo M, McInnes J, Nguyen HTT, Nevoigt E. 2013. Re-evaluation of glycerol utilization in Saccharomyces cerevisiae: characterization of an isolate that grows on glycerol without supporting supplements. Biotechnol Biofuels 6:157.

79. Lucas C, Da Costa M, Van Uden N. 1990. Osmoregulatory active sodium-glycerol co-transport in the halotolerant yeastDebaryomyces hansenii. Yeast 6:187–191.

80. Lages F, Lucas C. 1995. Characterization of a glycerol/H+ symport in the halotolerant yeast Pichia sorbitophila. Yeast 11:111–119.

81. Holst B, Lunde C, Lages F, Oliveira R, Lucas C, Kielland-Brandt MC. 2000. GUP1 and its close homologue GUP2, encoding multimembrane-spanning proteins involved in active glycerol uptake in Saccharomyces cerevisiae. Mol Microbiol 37:108–24.

82. Oliveira R, Lucas C. 2004. Expression studies of GUP1 and GUP2, genes involved in glycerol active transport in Saccharomyces cerevisiae, using semi-quantitative RT-PCR. Curr Genet 46:140–6.

83. Dušková M, Ferreira C, Lucas C, Sychrová H. 2015. Two glycerol uptake systems contribute to the high osmotolerance of Zygosaccharomyces rouxii. Mol Microbiol 97:541–59.

84. Liu X, Mortensen UH, Workman M. 2013. Expression and functional studies of genes involved in transport and metabolism of glycerol in Pachysolen tannophilus. Microb Cell Fact 12:27.

85. Kayikci O, Nielsen J. 2015. Glucose repression in Saccharomyces cerevisiaë. FEMS Yeast Res 15.

86. De Deken RH. 1966. The Crabtree Effect: A Regulatory System in Yeast. J Gen Microbiol 44:149–156.

87. Brauer MJ, Saldanha AJ, Dolinski K, Botstein D. 2005. Homeostatic adjustment and metabolic remodeling in glucose-limited yeast cultures. Mol Biol Cell 16:2503–17.

88. DeRisi JL, Iyer VR, Brown PO. 1997. Exploring the metabolic and genetic control of gene expression on a genomic scale. Science 278:680–6.

89. Roberts GG, Hudson AP. 2006. Transcriptome profiling of Saccharomyces cerevisiae during a transition from fermentative to glycerol-based respiratory growth reveals extensive metabolic and structural remodeling. Mol Genet Genomics 276:170–186.

90. Schüller H-J. 2003. Transcriptional control of nonfermentative metabolism in the yeast Saccharomyces cerevisiae. Curr Genet 43:139–60.

91. Grauslund M, Lopes JM, Rønnow B. 1999. Expression of GUT1, which encodes glycerol kinase in Saccharomyces cerevisiae, is controlled by the positive regulators Adr1p, Ino2p and Ino4p and the negative regulator Opi1p in a carbon source-dependent fashion. Nucleic Acids Res 27:4391–4398.

92. Grauslund M, Rønnow B. 2000. Carbon source-dependent transcriptional regulation of the mitochondrial glycerol-3-phosphate dehydrogenase gene, GUT2, from Saccharomyces cerevisiae. Can J Microbiol

47

46:1096–100.

93. Haurie V, Perrot M, Mini T, Jenö P, Sagliocco F, Boucherie H. 2001. The Transcriptional Activator Cat8p Provides a Major Contribution to the Reprogramming of Carbon Metabolism during the Diauxic Shift in Saccharomyces cerevisiae. J Biol Chem 276:76–85.

94. Lu L, Roberts GG, Oszust C, Hudson AP. 2005. The YJR127C/ZMS1 gene product is involved in glycerol-based respiratory growth of the yeast Saccharomyces cerevisiae. Curr Genet 48:235–246.

95. Lu L, Roberts G, Simon K, Yu J, Hudson AP. 2003. Rsf1p, a protein required for respiratory growth of Saccharomyces cerevisiae. Curr Genet 43:263–272.

96. Roberts GG, Hudson AP. 2009. Rsf1p is required for an efficient metabolic shift from fermentative to glycerol-based respiratory growth in S. cerevisiae. Yeast 26:95–110.

97. Kavšček M, Bhutada G, Madl T, Natter K. 2015. Optimization of lipid production with a genome-scale model of Yarrowia lipolytica. BMC Syst Biol 9:72.

98. Yuzbasheva EY, Mostova EB, Andreeva NI, Yuzbashev T V., Fedorov AS, Konova IA, Sineoky SP. 2018. A metabolic engineering strategy for producing free fatty acids by the Yarrowia lipolytica yeast based on impairment of glycerol metabolism. Biotechnol Bioeng 115:433–443.

99. Papanikolaou S, Muniglia L, Chevalot I, Aggelis G, Marc I. 2002. Yarrowia lipolytica as a potential producer of citric acid from raw glycerol. J Appl Microbiol 92:737–744.

100. Papanikolaou S, Rontou M, Belka A, Athenaki M, Gardeli C, Mallouchos A, Kalantzi O, Koutinas AA, Kookos IK, Zeng A-P, Aggelis G. 2017. Conversion of biodiesel-derived glycerol into biotechnological products of industrial significance by yeast and fungal strains. Eng Life Sci 17:262–281.

101. Workman M, Holt P, Thykaer J. 2013. Comparing cellular performance of Yarrowia lipolytica during growth on glucose and glycerol in submerged cultivations. AMB Express 3:58.

102. Iida T, Ohta A, Takagi M. 1998. Cloning and characterization of ann-alkane-inducible cytochrome P450 gene essential forn-decane assimilation byYarrowia lipolytica. Yeast 14:1387–1397.

103. Iida T, Sumita T, Ohta A, Takagi M. 2000. The cytochrome P450ALK multigene family of ann-alkane-assimilating yeast,Yarrowia lipolytica: cloning and characterization of genes coding for new CYP52 family members. Yeast 16:1077–1087.

104. Sherwood KE, Cano DJ, Maupin-Furlow JA. 2009. Glycerol-Mediated Repression of Glucose Metabolism and Glycerol Kinase as the Sole Route of Glycerol Catabolism in the Haloarchaeon Haloferax volcanii. J Bacteriol 191:4307–4315.

105. Ryu S, Trinh CT, Elliot MA. 2018. Understanding Functional Roles of Native Pentose-Specific Transporters for Activating Dormant Pentose Metabolism in Yarrowia lipolytica.

106. Seiboth B, Metz B. 2011. Fungal arabinan and L-arabinose metabolism. Appl Microbiol Biotechnol 89:1665–73.

107. Tantirungkij M, Izuishi T, Seki T, Yoshida T. 1994. Fed-batch fermentation of xylose by a fast-growing mutant of xylose-assimilating recombinant Saccharomyces cerevisiae. Appl Microbiol Biotechnol 41:8–12.

48

108. Kötter P, Amore R, Hollenberg CP, Ciriacy M. 1990. Isolation and characterization of the Pichia stipitis xylitol dehydrogenase gene, XYL2, and construction of a xylose-utilizing Saccharomyces cerevisiae transformant. Curr Genet 18:493–500.

109. Tsigie YA, Wang C-Y, Truong C-T, Ju Y-H. 2011. Lipid production from Yarrowia lipolytica Po1g grown in sugarcane bagasse hydrolysate. Bioresour Technol 102:9216–22.

110. Ledesma-Amaro R, Lazar Z, Rakicka M, Guo Z, Fouchard F, Coq A-MC-L, Nicaud J-M. 2016. Metabolic engineering of Yarrowia lipolytica to produce chemicals and fuels from xylose. Metab Eng 38:115–124.

111. Li H, Alper HS. 2016. Enabling xylose utilization in Yarrowia lipolytica for lipid production. Biotechnol J 11:1230–1240.

112. Rodriguez GM, Hussain MS, Gambill L, Gao D, Yaguchi A, Blenner M. 2016. Engineering xylose utilization in Yarrowia lipolytica by understanding its cryptic xylose pathway. Biotechnol Biofuels 9:149.

113. Ryu S, Hipp J, Trinh CT. 2015. Activating and Elucidating Metabolism of Complex Sugars in Yarrowia lipolytica. Appl Environ Microbiol 82:1334–45.

114. Rintala E, Wiebe MG, Tamminen A, Ruohonen L, Penttilä M. 2008. Transcription of hexose transporters of Saccharomyces cerevisiae is affected by change in oxygen provision. BMC Microbiol 8:53.

115. Lazar Z, Neuvéglise C, Rossignol T, Devillers H, Morin N, Robak M, Nicaud J-M, Crutz-Le Coq A-M. 2017. Characterization of hexose transporters in Yarrowia lipolytica reveals new groups of Sugar Porters involved in yeast growth. Fungal Genet Biol 100:1–12.

116. Workman M, Andersen MR, Thykaer J. 2013. Integrated Approaches for Assessment of Cellular Performance in Industrially Relevant Filamentous Fungi. Ind Biotechnol 9:337–344.

117. Harms P, Kostov Y, Rao G. 2002. Bioprocess monitoring. Curr Opin Biotechnol 13:124–7.

118. Sonnleitner B, Locher G, Fiechter A. 1992. Biomass determination. J Biotechnol 25:5–22.

119. Posch AE, Spadiut O, Herwig C. 2012. Switching industrial production processes from complex to defined media: method development and case study using the example of Penicillium chrysogenum. Microb Cell Fact 11:88.

120. Formenti LR, Nørregaard A, Bolic A, Hernandez DQ, Hagemann T, Heins A-L, Larsson H, Mears L, Mauricio-Iglesias M, Krühne U, Gernaey K V. 2014. Challenges in industrial fermentation technology research. Biotechnol J 9:727–738.

121. Neubauer P, Cruz N, Glauche F, Junne S, Knepper A, Raven M. 2013. Consistent development of bioprocesses from microliter cultures to the industrial scale. Eng Life Sci 13:224–238.

122. McNeil B, Harvey LM, Wiley InterScience (Online service). 2008. Practical fermentation technology. Wiley.

123. Yamanè T, Shimizu S. 1984. Fed-batch techniques in microbial processes, p. 147–194. In Bioprocess Parameter Control. Springer-Verlag, Berlin/Heidelberg.

124. Hakkaart XD V, Pronk JT, van Maris AJA. 2017. A Simulator-Assisted Workshop for Teaching Chemostat Cultivation in Academic Classes on Microbial Physiology. J Microbiol Biol Educ 18.

125. Simpson JT, Pop M. 2015. The Theory and Practice of Genome Sequence Assembly. Annu Rev Genomics

49

Hum Genet 16:153–172.

126. Ekblom R, Wolf JBW. 2014. A field guide to whole-genome sequencing, assembly and annotation. Evol Appl 7:1026–1042.

127. Fleischmann RD, Adams MD, White O, Clayton RA, Kirkness EF, Kerlavage AR, Bult CJ, Tomb JF, Dougherty BA, Merrick JM. 1995. Whole-genome random sequencing and assembly of Haemophilus influenzae Rd. Science 269:496–512.

128. Goffeau A, Barrell BG, Bussey H, Davis RW, Dujon B, Feldmann H, Galibert F, Hoheisel JD, Jacq C, Johnston M, Louis EJ, Mewes HW, Murakami Y, Philippsen P, Tettelin H, Oliver SG. 1996. Life with 6000 genes. Science 274:546, 563–7.

129. Reuter JA, Spacek D V, Snyder MP. 2015. High-throughput sequencing technologies. Mol Cell 58:586–97.

130. Andrews S. 2010. FastQC: a quality control tool for high throughput sequence data. Available online at: http://www.bioinformatics.babraham.ac.uk/projects/fastqc.

131. Schubert M, Lindgreen S, Orlando L. 2016. AdapterRemoval v2: rapid adapter trimming, identification, and read merging. BMC Res Notes 9:88.

132. Proux-Wéra E, Armisén D, Byrne KP, Wolfe KH. 2012. A pipeline for automated annotation of yeast genome sequences by a conserved-synteny approach. BMC Bioinformatics 13:237.

133. Reid I, O’Toole N, Zabaneh O, Nourzadeh R, Dahdouli M, Abdellateef M, Gordon PM, Soh J, Butler G, Sensen CW, Tsang A. 2014. SnowyOwl: accurate prediction of fungal genes by using RNA-Seq and homology information to select among ab initio models. BMC Bioinformatics 15:229.

134. Nookaew I, Papini M, Pornputtapong N, Scalcinati G, Fagerberg L, Uhlén M, Nielsen J. 2012. A comprehensive comparison of RNA-Seq-based transcriptome analysis from reads to differential gene expression and cross-comparison with microarrays: a case study in Saccharomyces cerevisiae. Nucleic Acids Res 40:10084–10097.

135. Wilhelm BT, Landry J-R. 2009. RNA-Seq—quantitative measurement of expression through massively parallel RNA-sequencing. Methods 48:249–257.

136. Jewett MC, Oliveira AP, Patil KR, Nielsen J. 2005. The role of high-throughput transcriptome analysis in metabolic engineering. Biotechnol Bioprocess Eng 10:385–399.

137. Conesa A, Madrigal P, Tarazona S, Gomez-Cabrero D, Cervera A, McPherson A, Szcześniak MW, Gaffney DJ, Elo LL, Zhang X, Mortazavi A. 2016. A survey of best practices for RNA-seq data analysis. Genome Biol 17:13.

138. Liao Y, Smyth GK, Shi W. 2014. featureCounts: an efficient general purpose program for assigning sequence reads to genomic features. Bioinformatics 30:923–930.

139. Law CW, Chen Y, Shi W, Smyth GK. 2014. voom: precision weights unlock linear model analysis tools for RNA-seq read counts. Genome Biol 15:R29.

140. Love MI, Anders S, Kim V, Huber W. 2015. RNA-Seq workflow: gene-level exploratory analysis and differential expression. F1000Research 4:1070.

50

141. Law CW, Alhamdoosh M, Su S, Smyth GK, Ritchie ME. 2016. RNA-seq analysis is easy as 1-2-3 with limma, Glimma and edgeR. F1000Research 5:1408.

142. Anders S, Huber W. 2010. Differential expression analysis for sequence count data. Genome Biol 11:R106.

143. Love MI, Huber W, Anders S. 2014. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol 15:550.

144. Robinson MD, McCarthy DJ, Smyth GK. 2010. edgeR: a Bioconductor package for differential expression analysis of digital gene expression data. Bioinformatics 26:139–40.

145. Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, Smyth GK. 2015. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res 43:e47–e47.

146. Robinson MD, Oshlack A. 2010. A scaling normalization method for differential expression analysis of RNA-seq data. Genome Biol 11:R25.

147. Väremo L, Nielsen J, Nookaew I. 2013. Enriching the gene set analysis of genome-wide data by incorporating directionality of gene expression and combining statistical hypotheses and methods. Nucleic Acids Res 41:4378–91.

51

Part II: Manuscripts

52

Manuscript 1: Physiological comparison of Yarrowia 1

lipolytica strains reveals differences in the utilization 2

of sugars and glycerol 3

4

Patrice Lubutaa, Christopher T. Workman#a and Mhairi Workmana* 5

aDepartment of Biotechnology and Biomedicine, Technical University of Denmark, Kgs. Lyngby, Denmark 6

*Present address: Mhairi Workman, Novo Nordisk, Bagsværd, Denmark. 7

8

9

Running Head: Physiological comparison of Y. lipolytica strains 10

11

#Address correspondence to Christopher T. Workman, cwor@dtu.dk. 12

13

Keywords: Yarrowia lipolytica, quantitative physiology, glycerol, glucose, xylose, arabinose 14

15

16

17

18

53

Abstract 19

Yarrowia lipolytica is a non-conventional yeast species with a high potential for novel and sustainable 20

biotechnological applications. Here we present a quantitative comparison of physiology between Y. lipolytica 21

IBT 446, a Danish feta cheese isolate, and the frequently used wild type strains W29 and H222, also known as 22

the French and German strain respectively. The physiology was assessed in single- and mixed-carbon cultivations 23

using sugars (glucose, xylose and arabinose) and glycerol as carbon sources. Inter-strain variations were detected 24

and, in particular, IBT 446 was found to differ in several characteristics from the commonly used strains. In single 25

substrate experiments glycerol was the preferred carbon source for all three strains, and IBT 446 showed the 26

highest yield of sugar alcohols (primarily mannitol). The strains displayed sequential substrate utilization in 27

mixed-carbon conditions but only IBT 446 was able to utilize all four substrates. W29 and H222 did not use 28

arabinose, while xylose consumption ended after approximately 50 % depletion. Xylose was converted into the 29

valuable sugar alcohol xylitol by all three strains, although only IBT 446 was able to subsequently utilize xylitol. 30

Furthermore, co-consumption of glycerol and glucose was observed to vary between the strains, indicating strain 31

specific carbon source regulation. 32

Importance 33

The nonconventional yeast species Y. lipolytica has gathered attention as a promising cell factory during the last 34

years. Various strains have been applied by the research community and differences in substrate utilization and 35

product formation have been observed. Here, we compared the physiology of the Technical University of 36

Denmark´s in-house strain Y. lipolytica IBT 446, with the commonly used wild type strains W29 and H222. 37

Evidence for a natural strain diversity could be provided by highly-controlled bioreactor experiments and the use 38

of defined minimal media. IBT 446 should be considered as a new host for bioprocess applications due to several 39

physiological benefits such as the lack of hyphae formation, polyol yield and pentose consumption. Furthermore, 40

54

this study demonstrates a glycerol repression-like effect, which is strain specific and in accordance with previous 41

reports. This paves the way for future studies analyzing novel regulatory mechanism in this species. 42

Introduction 43

The non-conventional yeast species Yarrowia lipolytica, is known to grow on a diverse range of substrates 44

including hydrophobic substances, C6-sugars, alcohols and acetate (1–4). Y. lipolytica is also naturally able to 45

synthesize various economically relevant products including lipids (to levels over 50 % of cell dry weight), sugar 46

alcohols (e.g. mannitol, erythritol), organic acids (e.g. citric acid, isocitric acid and α-ketoglutaric acid) and 47

exoenzymes (proteases and lipases) (1, 5, 6). 48

During the last years, this species gathered attention as a promising new host for biotechnological applications. 49

Many different strains have been applied by the Yarrowia community and differences in substrate utilization and 50

product formation have been observed (7–9). Comparison of the physiological studies is complicated by the use 51

of growth supporting supplements (e.g. complex media components or amino acid mixes), as well as cultivation 52

techniques with poor reproducibility, lacking control and monitoring of important cultivation parameters. 53

Here we report on Y. lipolytica IBT 446, the Technical University of Denmark´s in-house strain, originally isolated 54

from feta cheese (10). The physiology of this strain was assessed previously demonstrating a fast growth on 55

glycerol, a clear preference for glycerol over glucose and the production of sugar alcohols (11, 12). 56

Heterologously expressed aquaglyceroporin Fps1 homologues from IBT 446 also improved glycerol uptake in 57

S. cerevisiae providing insights into glycerol transport mechanisms (12). The aim of this study was to compare 58

the physiology of the Danish strain IBT 446 with the frequently used French strain W29 and German strain H222. 59

Controlled bioreactor-experiments and well-defined medium compositions were applied in order to investigate 60

Y. lipolytica´s natural strain diversity. As benchmarking conditions, glycerol and the sugars glucose, xylose and 61

arabinose were chosen. 62

55

Results 63

Benchmarking strains in single substrate cultivations 64

In initial bioreactor batch experiments, the Y. lipolytica strains W29 and H222 were grown in minimal medium 65

containing 0.65 cmole L-1 (≈ 20 g L-1) of either glycerol or glucose as the sole carbon source. Concentrations based 66

on cmole were chosen to provide the cells with the same amount of carbon despite using carbon sources with 67

different amounts of C-atoms (see Materials and Methods Table 3). These experiments were carried out in order 68

to benchmark the commonly used strains W29 and H222 with Y. lipolytica strain IBT 446 which was investigated 69

previous by Workman et al. (2013) under the same experimental conditions. The cultivation profiles of W29 and 70

H222 on glycerol and glucose can be found in Fig. S1/S2 and of IBT 446 in the mentioned publication. Table 1 71

summarizes the physiological parameters of the three strains in single substrate cultivations. Growth rates were 72

higher for all strains grown in glycerol containing media compared to glucose containing media: 0.30 h-1 (IBT 73

446), 0.31 h-1 (W29), 0.35 h-1 (H222) on glycerol compared to 0.24 h-1 (IBT), 0.28 h-1 (W29), 0.28 h-1 (H222) on 74

glucose. The exponential growth phase ceased when oxygen became limiting, indicated by a dissolved oxygen 75

level of 0 %. The Y. lipolytica strains exhibited linear growth from this time point onwards. 76

The strains W29 and H222 produced only biomass and CO2 when glucose was the sole carbon source. In contrast, 77

when grown on glycerol, both strains produced small amounts of the polyol mannitol. Yield coefficients revealed 78

that all three strains behaved similarly when grown on glucose. Approximately 65-70 % of carbon went into 79

biomass and around 30 % into CO2 production. In contrast, strain differences were observable under glycerol 80

conditions: IBT 446 produced more polyols than W29 and H222. Yield coefficients revealed that in the IBT 446 81

cultivations 17 % of the available carbon went into polyol production, 61 % into biomass and 23 % into CO2. For 82

W29 more carbon was used for biomass (73 %) and CO2 production (30 %), while only ≈ 1.5 % went into polyol 83

production. 84

56

Table 1: Physiological parameters of single carbon cultivations of Y. lipolytica IBT 446, W29 and H222 on glucose and on 85 glycerol. 86

IBT 4461 W29 H222

Glucose Glycerol Glucose Glycerol Glucose Glycerol

Growth rate

μmax (h-1) 0,24 ± 0,01 0,30 ± 0,01 0.28 ± 0.02 0.31 ± 0.02 0.28 ± 0.01 0.35 ± 0.01

Yield coefficients

Ysx (cmole cmole-1) 0,69 ± 0,03 0,61 ± 0,01 0,67 ± 0,03 0,73 ± 0,01 0,64 ± 0,02 0.72 ± 0,01

Ysc (cmole cmole-1) 0,30 ± 0,01 0,23 ± 0,00 0,34 ± 0,02 0,30 ± 0,01 0,35 ± 0,004 0,21 ± 0,01

Ysm (cmole cmole-1) N/A* 0,17 ± 0,00 N/A* 0,015 ± 0,01 N/A* N/A**

Total 0,99 ± 0,03 1,01 ± 0,01 1,02 ± 0,05 1,04 ± 0,01 0,98 ± 0,01 0,94 ± 0,01

1Results from (Workman et al. 2013). N/A: Not applicable: *below detection limit. **Carbon source not depleted at last sample point.

87

To the best of our knowledge, contrary to W29 and its derivatives, there are no studies available investigating 88

the ability of IBT 446 and H222 to grow on xylose or arabinose as the sole carbon source. Therefore, a growth 89

screen in defined minimal media and the absence of complex media components was conducted. The cultivation 90

was performed in shake flasks with 0.65 cmole L-1 (≈ 20 g L-1) of each pentose, and continued for 140 h. No 91

biomass accumulation could be detected for any of the strains and HPLC analysis showed no consumption of the 92

substrates (data not shown). The results are confirming previous findings that Y. lipolytica is not able to utilize 93

pentose sugars as the sole source of carbon and energy in minimal media conditions (see below). 94

Multiple carbon cultivations 95

Mixed carbon cultivations were performed in order to compare the cellular response of each strain to multiple 96

carbon sources present in the environment. Therefore, IBT 446, W29 and H222 were cultivated in bioreactors 97

with equal cmole amounts of glycerol, glucose, xylose and arabinose (each 0.163 cmole L-1 ≈ 5 g L-1) giving a total 98

amount of 0.65 cmole L-1 ≈ 20 g L-1 (see Materials and Methods Table 3). The cultivations were carried out in 99

triplicates and continued for several days until carbon depletion or no changes could be detected anymore: 86 h 100

57

(IBT 446), 105 h (W29) and 87 h (H222). Replicates were averaged and results for the first 40 hours are shown in 101

Fig. 1. The full cultivation profiles for each strain can be found in the supplemental material (Fig. S3-S5). 102

After an initial lag phase, a sequential carbon source utilization pattern was observed for all strains, although IBT 103

446 was the only strain which was able to utilize all four substrates (Fig. 1 E). This strain used the carbon sources 104

in the order: 1. glycerol, 2. glucose, 3. xylose and 4. arabinose. Arabinose consumption was slower compared to 105

the other substrates. W29 and H222 did not consume arabinose, and xylose utilization stopped after 106

approximately 50 % of the available xylose was used. Nevertheless, the general order of sequential substrate 107

utilization was the same as for IBT 446 (Fig. 1 A and C). 108

Glycerol was the preferred substrate for all three strains and was depleted first, followed by glucose. 109

Interestingly, the degree of co-consumption of glycerol and glucose differed between the strains. IBT 446 110

consumed glucose only in small amounts when glycerol was available, then rapidly increased glucose utilization 111

when glycerol was depleted. In IBT 446 fermentations, an approximate 3-hour time difference was observed 112

between glycerol and glucose depletion. In contrast, W29 co-fermented glycerol and glucose nearly 113

simultaneously, displaying less than 0.5 h between the depletion of glycerol and glucose. The carbon utilization 114

of H222 was more similar to that of IBT 446 although with somewhat less time difference between glycerol and 115

glucose depletion (2.4 h difference). 116

58

117

Fig. 1 Mixed carbon cultivations of different Y. lipolytica strains. Upper panel: Substrate and product 118

concentrations (cmole L-1) of W29 (A), H222 (C) and IBT 446 (E) cultivations. Bottom panel: Biomass concentration 119

(cmole L-1), accumulated CO2 (cmole L-1), and dissolved oxygen level in the broth (%) of W29 (B), H222 (D) and 120

IBT 446 (F) cultivations. The O2 molar ratio of the exhaust gas is also shown. 121

The cultivations of all three strains became oxygen limited already during growth on glycerol, indicated by the 122

measurement of negligible levels of dissolved oxygen. Y. lipolytica is a strictly respiratory yeast with a high 123

demand for oxygen (13). It was shown previously, that the reducting force in Y. lipolytica is provided by the 124

pentose phosphate pathway and that the enzyme transketolase is a crucial enzyme for growth under oxygen 125

limitation (14). Interestingly, also the dissolved oxygen profile differed between the strains. The dissolved oxygen 126

level in strain W29 cultivations remained low until also glucose was consumed (Fig. 1 B). The dissolved oxygen 127

level then increased briefly after glucose depletion. By contrast, cultivations with IBT 446 and H222 displayed a 128

period of low respiration after glycerol depletion and before glucose utilization, indicated by sudden increase of 129

the dissolved oxygen level (Fig. 1 D and F). Dissolved oxygen level decreased again while the strains started to 130

59

utilize glucose. This effect was the strongest in IBT 446 cultivations and appearing to a lesser extent in H222 131

cultivations. The dissolved oxygen profile was additionally confirmed by off gas analysis measuring the oxygen 132

molar ratio in the exhaust gas. 133

All three strains were able to utilize xylose after glucose depletion. Compared to single carbon experiments with 134

xylose, the three strains were able to utilize this substrate in the presence of other carbon sources. This is in 135

accordance with previous literature where xylose and glucose were co-utilized (2, 15). Interestingly, IBT 446 was 136

able to utilize all available xylose whereas W29 an H222 consumed only about 50 %. All three strains produced 137

the sugar alcohol xylitol during the cultivation, which was measurable in the supernatant indicating functional 138

xylose transport and xylose reductase (XYL1) activity. The production started during oxygen limitation, when the 139

primary carbon sources (glycerol and glucose) were nearly depleted and conversion of xylose began. For W29 140

and H222 xylitol concentration stayed constant after its extracellular accumulation, whereas the xylitol level 141

decreased again for strain IBT 446, indicating a re-consumption of this product. The maximum xylitol 142

concentrations of 0.05 cmole L-1 ≈ 1.5 g L-1 (H222), 0.06 cmole L-1 ≈ 1.8 g L-1 (W29) and 0.09 cmole L-1 ≈ 2.7 g L-1 143

(IBT 446) were obtained. Additionally, the strains W29 and H222 also produced small amounts of mannitol (max 144

0.004 cmole L-1 ≈ 0.1 g L-1 each). IBT 446 was the only strain capable of utilizing arabinose although the 145

consumption rate was much slower than that for xylose. However, after 86 h, the arabinose concentration 146

decreased from initial 0.17 cmole L-1 ≈ 5.1 g L-1 to 0.03 cmole L-1 ≈ 0.9 g L-1, which is a consumption of 147

approximately 75 % of the available pentose sugar (Fig. S3). 148

Growth rates of IBT 446, W29 and H222 in the mixed carbon cultivation were 0.32, 0.38, and 0.37 respectively. 149

The three strains grew slightly faster compared to the cultivations performed on glycerol as the sole carbon 150

source. Table 2 summarizes the yield coefficients and growth rates of the three strains in the mixed carbon 151

cultivation. IBT 446 had a lower yield of biomass on total carbon (0.43 ± 0.02 cmole cmole-1) compared W29 (0.59 152

60

± 0.04 cmole cmole-1) and H222 (0.59 ± 0.02 cmole cmole-1), while the carbon dioxide yield (Ysc) was similar for 153

all strains. 154

Table 2: Physiological parameters of mixed carbon cultivations of Y. lipolytica IBT 446, W29 and H222. 155

IBT 446 W29 H222

glycerol + glucose + xylose + arabinose

Growth rate

μmax (h-1) 0.32 ± 0.03 0.38 ± 0.01 0.37 ± 0.01

Yield coefficients

Ysx (cmole cmole-1) 0.43 ± 0.02 0.59 ± 0.04 0.59 ± 0.02

Ysc (cmole cmole-1) 0.39 ± 0.03 0.39 ± 0.03 0.35 ± 0.01

Ysxy (cmole cmole-1) 0.13 ± 0.005 0.11 ± 0.01 0.09 ± 0.01

Ysm (cmole cmole-1) N/A* 0.01 ± 0.005 0.01 ± 0.002

Total 0.95 ± 0.04 overall 1.09 ± 0.05 overall 1.1 ± 0.05

N/A: Not applicable: *below detection limit

156

Biomass accumulation of all three strains took place only during growth on glycerol and glucose. Therefore, these 157

substrates can be seen as primary carbon sources. In contrast, xylose and arabinose are only utilized in the 158

presence of these primary carbon sources, where they can be seen as secondary substrates. Interestingly, the 159

biomass concentration of IBT 446 stayed constant after depletion of the primary carbon sources (Fig. S3) while 160

it decreased for W29 (Fig. S4) and H222 (Fig. S5). The cells of IBT 446 were metabolically active throughout the 161

whole cultivation period, indicated by the continued production of CO2 and the consumption of xylose, xylitol 162

and arabinose until the cultivation was stopped. By contrast, H222 did not consume xylose, xylitol or arabinose 163

and CO2 was only slightly produced. After depletion of the primary carbon sources, W29 showed a similar 164

behavior to H222. 165

Interestingly, we did not experienced any hyphae formation of IBT 446 under the tested conditions so far. In 166

contrast, hyphae formation was usually detected for W29 and H222. It was previously reported that filamentous 167

61

growth is triggered in media containing glucose, while the yeast form is predominant on glycerol (16). We 168

evaluated the morphology of the three strains on YPD and YPG agar plates and documented the cell morphology 169

(data not shown). In accordance to previous observations, IBT 446 appeared only in the yeast form, whereas the 170

other two strains exhibited hyphae formation. 171

Discussion 172

Research of Y. lipolytica is characterized by a diversification of the strains applied by the Yarrowia-community 173

and cases of strain variation have been reported (7, 8). This study provides a physiological comparison between 174

the feta cheese isolate Y. lipolytica IBT 446 and the two frequently used linages W29 and H222, isolated from 175

sewage water and soil respectively, in an highly controlled experimental setup (information: CIRM-Levures strain 176

catalogue). The Danish strain IBT 446 possesses beneficial properties like the lack of hyphae formation in all 177

tested conditions so far. Hyphae formation is problematic for industrial applications and strains unable to 178

undergo yeast-to-hyphae formation are desired (17). 179

In initial single-substrate experiments, W29 and H222 were characterized and compared with previous results 180

from IBT 446 obtained by Workman et al. (2013) under identical conditions. All strains showed a higher growth 181

rate on glycerol than on glucose, which is in accordance with several previous studies (11, 18, 19). When grown 182

on glucose, all strains produced only biomass and CO2. In contrast, polyols accumulated in the supernatant when 183

grown on glycerol. It has been shown previously, that Y. lipolytica produces polyols as a response to osmotic 184

pressure (20). In this study, carbon source concentrations were adjusted based on cmole instead of mole to 185

increase comparability between substrates with varying carbon atoms. The same cmole amount (0.65 cmole L-1) 186

is equivalent to twice the molar glycerol amount (0.217 mole L-1) compared to moles of glucose (0.108 mole L-1). 187

The higher molarity in glycerol fermentations potentially triggered the polyol synthesis. Polyol production was 188

also reported to be strain dependent (21), and since IBT 446 showed a higher polyol yield than W29 and H222, 189

this strain should be further assessed for its capacity to produce these economically relevant compounds. 190

62

Mixed substrate experiments were performed in order to compare the strain´s metabolic response to multiple 191

carbon sources present in the environment. Cultivation profiles revealed that the three strains consumed the 192

substrates in a sequential manner. IBT 446 was the only strain able to consume all four carbon sources whereas 193

W29 and H222 did not consume arabinose and only approximately 50 % of the available xylose. The general 194

order of substrate utilization was: 1. glycerol, 2. glucose, 3. xylose and 4. arabinose (only IBT 446). The preference 195

for glycerol over glucose in co-substrate cultivations has been reported previously for several Y. lipolytica strains 196

(11, 22, 23). Interestingly, the degree of glycerol-glucose co-consumption varied between the strains. W29 197

converted the two substrates nearly simultaneously, whereas IBT 446 and H222 converted glucose only in small 198

amounts when glycerol was still present. Strain-to-strain variation of substrate co-utilization has been also 199

reported previously (23, 24), indicating transcriptional or biochemical differences. The delayed glucose 200

consumption in the presence of glycerol, which results in a 3-hour time difference in depletion for IBT 446, 201

implies a glycerol repressive effect on the utilization of glucose in these strains. This effect seems to be nearly 202

absent in W29, leading to a high degree of simultaneous utilization. This observation was supported by profiles 203

of dissolved oxygen and exhaust gas composition. A low respiration rate transition period was observed in IBT 204

446 and H222 cultivations, which indicated a diauxic shift where metabolism was adjusting to the utilization of 205

glucose. In contrast to known carbon catabolite repression (CCR) mechanisms, ensuring the preferred use of 206

glucose, reports about glycerol repression mechanisms are rare. Only one study by Sherwood et al. (2009) 207

reported a repression of glucose uptake by glycerol in the archaeon Haloferax volcanii. Several studies with 208

Y. lipolytica, however, reported transcriptional repression of genes involved in n-alkane assimilation in the 209

presence of glycerol (19, 26, 27). Interestingly, in total contrast to observations in the present study, Yuzbasheva 210

et al. (2018) reported about a W29 strain exhibiting a strong glycerol repression on the utilization of glucose 211

during co-substrate cultivations. Further investigations addressing this potential glycerol repressive effect are 212

necessary, and might uncover so far unknown regulatory mechanisms in yeasts. 213

63

None of the strains grew on xylose when applied as the sole carbon source, but the strains were able to use this 214

substrate in combination with glycerol and glucose in the mixed carbon cultivations. The consumption started 215

close to the time point of glucose depletion. IBT 446 was the only strain consuming all xylose, while W29 and 216

H222 stopped the consumption after approximately 50 % xylose depletion. Further, all three strains converted 217

xylose into the intermediate xylitol, which accumulated in the supernatant, and only IBT 446 was able to re-218

utilized xylitol again. These findings are in accordance with previously published results: Until now, studies 219

addressing xylose utilization in Y. lipolytica included only W29 and derivatives thereof, such as Po1d, PO1f, Po1g, 220

Po1t or E26 (2, 28–32). Here, we tested for the first time xylose consumption in the strains IBT 446 and H222, 221

however, it seems that unmodified Y. lipolytica strains either do not or only grow on xylose after an adaptive 222

evolution approach (adaption phase) was performed (2). Only one study observed proper growth of Y. lipolytica 223

on xylose as a sole carbon source, but in this case complex media components (peptone and yeast extract) were 224

included in the media, providing potentially growth benefits (32). It was shown previously that Y. lipolytica 225

possess an endogenous oxidoreductase catalytic pathway for the utilization of xylose, including the enzymes 226

xylose reductase (XYL1), xylitol dehydrogenase (XYL2) and xylulose kinase (XYL3) (33). This pathway, however, 227

appears to be predominantly cryptic, since transcriptional activation of the involved genes is insufficient (30). As 228

in the present study, xylitol accumulation was reported, and the authors confirmed that the conversion to 229

xylulose, catalyzed by the enzyme xylitol dehydrogenase (XYL2), was a limiting step (2). Current engineering 230

efforts, therefore, are focused on the overexpression of endogenous or heterologous xyl1-3 catabolic genes in 231

Y. lipolytica (2, 15, 29, 34). 232

In the present study, glycerol and glucose functioned as primary carbon sources, since they led to the 233

accumulation of biomass exclusively. In contrast, pentoses were converted in the presence of these primary 234

carbon sources, however, this conversion did not appear to contribute to cell growth. The oxidoreductase 235

pathway is cofactor dependent (XYL1 and XYL2 enzymatic step) and cofactor imbalance has been described 236

64

previously for S. cerevisiae strains expressing an heterologous oxidoreductase pathway (35). Limitations in the 237

supply of cofactors could explain why xylose consumption stops suddenly in fermentations with W29 and H222 238

and additionally, why these strains were not able to consume the intermediate xylitol again. This is further 239

supported by the observation that, in contrast to W29 and H222, cells of IBT 446 were metabolically still active 240

after depletion of the primary carbon sources. In this period, IBT 446 re-utilized xylitol and also consumed 241

arabinose, while exhibiting respiration indicated by O2 and CO2 exhaust gas measurements. It seems that IBT 446 242

was the only strain, able to use the pentoses in order to keep its biomass-level constant (maintenance), while 243

the biomass concentration of W29 and H222 decreased directly after glycerol/glucose depletion. 244

Arabinose metabolism is less investigated in Y. lipolytica. In this study, arabinose consumption occurred only in 245

mixed substrate cultivations with IBT 446. One study exists in which arabinose transport and assimilation in 246

Y. lipolytica was investigated (36) and the authors suggested a putative arabinose catabolic pathway consisting 247

of the enzymes arabinose reductase (ARD), arabitol dehydrogenase (ADH), and xylulose reductase (XLR). 248

However, as in the case of xylose, arabinose utilization in Y. lipolytica seems to be limited due to insufficient 249

expression and potential cofactor imbalance. 250

Conclusion 251

The study demonstrates several physiological features which distinguish the feta cheese isolate IBT 446 from the 252

commonly used linages W29 and H222. Under all tested conditions so far, IBT 446 has only been observed to 253

grow in the yeast form, increasing its usability in fermentation settings. The polyol yield of this strains was high 254

and polyol production should be further assessed in future studies. This study has further provided quantitative 255

physiological evidence that the degree of glycerol-glucose co-utilization is strain dependent. Future studies 256

should address this glycerol repression-like effect, and can potentially reveal so far unknown regulation 257

mechanisms in yeasts. Finally, it has been demonstrated here that, as in the case of W29 and derivatives thereof, 258

pentose sugars cannot be used by IBT 446 and H222 when applied as the sole carbon source. However, IBT 446 259

65

was able to consume xylose, xylitol and arabinose in mixed substrate cultivations and use these substrates to 260

maintain biomass. 261

Materials and methods 262

Microorganisms 263

Three Y. lipolytica strains were used in this study. Y. lipolytica IBT 446 was obtained from the Department of 264

Biotechnology and Biomedicine’s culture collection, Technical University of Denmark. Y. lipolytica W29 (CLIB 89) 265

and Y. lipolytica H222 (CLIB 80) were obtained from CIRM-Levures strain collection, Institute National de la 266

Recherche Agronomique (INRA), France. For short-term storage, the strains were grown on YPD plates for 2 days 267

at 30 °C and the plates were stored at 4 °C. For long-term storage cells grown in YPD liquid media were kept at -268

80 °C in 17 % (w/w) glycerol. YPG (glycerol) plates were used to compare the morphology with YPD plates by 269

microscopy. 270

Cultivation medium 271

For all cultivations in this study defined minimal medium was used, containing chemicals of analytical grade: 272

(NH4)2SO4, 5.0 g L-1; KH2PO4, 3.0 g L-1, MgSO4.7H2O, 0.5 g L-1; Antifoam 298 (Sigma-Aldrich), 0.05 mL L-1; trace 273

metal solution, 1 mL L-1 (composed of: FeSO4.7H2O, 3 g L-1; ZnSO4.7H2O, 4.5 g L-1; CaCl2.6H2O, 4.5 g L-1; MnCl2.2H2O, 274

0.84 g L-1; CoCl2.6H2O, 0.3 g L-1; CuSO4.5H2O, 0.3 g L-1; Na2 MoO4.2H2O, 0.4 g L-1; H3BO3, 1 g L-1; KI, 0.1 g L-1; 275

Na2EDTA.2H2O, 15 g L-1) in deionized water. The pH was adjusted to 4.5 by NaOH prior to autoclavation. After 276

autoclavation 1 mL L-1 vitamin solution (composed of: d-biotin, 25 mg L-1; Ca-pantothenate, 0.5 g L-1; thiamin-277

HCl, 0.5 g L-1; pyridoxin-HCl, 0.5 g L-1; nicotinic acid, 0.5 g L-1; p-aminobenzoic acid, 0.1 g L-1; m-inositol, 12.5 g L-1) 278

was filter-sterilized (0.22 µm filter) and added with the separately autoclaved carbon source to the medium. 279

Different sets of cultivations were performed, always with a total carbon concentration of 0.65 cmole L-1 (≈ 20 g 280

L-1). In single carbon experiments 0.65 cmole L-1 of either glycerol, glucose, xylose or arabinose were used. In 281

mixed carbon cultivations equal amounts (0.163 cmole L-1 ≈ 5 g L-1) of glycerol, glucose, xylose and arabinose 282

66

were used resulting in a total concentration of 0.65 cmole L-1 carbon. The unit cmole L-1 was used to take the 283

varying amounts of C-atoms in the chemical formula of the carbon substrates into account (Table 3): 284

For example, glucose (C6H12O6) has double the cmole amount of glycerol (C3H8O3) and is also nearly double the 285

molecular weight (180.16 g mole-1 versus 92.09 g mole-1). Normalized to the number of carbon atoms, glucose 286

(CH2O) is 30.02 g cmole-1 and glycerol (CH2.66O) 30.7 g cmole-1. In this study, 0.65 cmole L-1 of glucose as well as 287

0.65 cmole L-1 of glycerol was used to provide the cells with the same amount of carbon. 288

Table 3. Carbon source concentrations used in the cultivation media. Concentrations were adjusted based on cmoles. 289 Different units (cmole L-1, g L-1, mole L-1) are shown to improve comparability. 290

Substrate Glucose Arabinose Xylose Glycerol

Mw (g mole-1) 180.16 150.13 150.13 92.09

chemical formula C6H12O6 C5H10O5 C5H10O5 C3H8O3

single carbon cultivations

concentration (cmole L-1) 0.65 0.65 0.65 0.65

concentration (g L-1) 19.52 19.52 19.52 19.95

concentration (mole L-1) 0.108 0.130 0.130 0.217

mixed carbon cultivations

concentration (cmole L-1) 0.163 0.163 0.163 0.163

concentration (g L-1) 4.89 4.89 4.89 5.00

concentration (mole L-1) 0.027 0.033 0.033 0.054

291

Inoculum preparation 292

Precultures for bioreactor cultivations were prepared by growing the strains in baffled shake flasks (500 ml) 293

containing 50 ml medium with the same carbon source as the intended batch cultivation. The incubation took 294

place at 30 °C and 150 rpm in a rotary shaker (Thermo Fisher Scientific). For the inoculation of the bioreactor 295

cultures, cells were harvested in mid exponential phase. 296

Precultures for shake flask cultivations were prepared as described above. As poor growth was expected on 5 g 297

L-1 xylose or 5 g L-1 arabinose as the sole carbon source, these pre cultures also contained 15 g L-1 glycerol. Before 298

67

inoculation of the main cultures, the cells were washed with the same media but lacking the carbon source to 299

remove residual glycerol. 300

Batch cultivations 301

Bioreactor based batch cultivations were performed in automated BIOSTAT® B fermenters with the working 302

volume of 2 L or in 1 L BIOSTAT® Q plus fermenters with 1 L working volume (both Sartorius Stedim Biotech S.A). 303

The following cultivation parameters were controlled and monitored: Temperature 30 °C +/- 1 °C; stirring rate 304

600 rpm; pH 4.5 +/− 0.1 by automatic addition of 2 M NaOH and 2 M sulphuric acid, aeration of 1 volume per 305

volume per minute (vvm) (1 standard liter per minute (slpm)) with atmospheric air. The gas analyzer Prima PRO 306

Process Mass Spectrometer (Thermo Fisher Scientific) was used for online measurement of exhaust carbon 307

dioxide and oxygen. Partial oxygen pressure (pO2) was constantly monitored with the electrochemical oxygen 308

sensor OxyFerm FDA 160 (Hamilton). Two kinds of batch experiments were performed: 1) single carbon 309

experiments in which Y. lipolytica strains W29 and H222 were grown in minimal media containing either glucose 310

or glycerol as the sole carbon source; 2) mixed substrate cultivations where Y. lipolytica IBT 446, W29 and H222 311

were grown in minimal media with equal concentrations of glycerol, glucose, xylose and arabinose. 312

Determination of biomass concentration 313

The biomass concentration was determined through spectrophotometry measurements and dry weight 314

determinations. The optical density of fermentation samples was measured at 600 nm on a UV-mini 1240 315

spectrophotometer (Shimadzu). For estimation of biomass dry weight a known volume of the fermentation broth 316

was filtered through pre-weighed 0.45 μm nitrocellulose filters (Sartorius Stedium) and dried in a microwave 317

oven at 150 W for 20 min. After cooling down in a desiccator the filters were weighed again. 318

Elemental analysis of Yarrowia lipolytica biomass 319

Elemental analysis (EA) of Y. lipolytica biomass was performed on a VARIO EL elemental analyzer (Elementar) 320

determining the percentage of carbon, hydrogen, nitrogen and sulfur. Oxygen percentage was assumed to make 321

68

up the remaining weight. Duplicate biomass samples were obtained from shake flask cultivations performed on 322

minimal media with the following conditions: IBT 446 on glycerol, IBT 446 on glucose, W29 on glycerol, W29 on 323

glucose, H222 on glycerol and H222 on glucose. Each biomass sample was measured in three technical replicates. 324

Table 4 shows the molecular formula and the molecular weight of Y. lipolytica biomass in different conditions. 325

Table 4. Elemental analysis of Y. lipolytica biomass under different growth conditions. 326

Glycerol Glucose

Molecular formula Molecular weight (g mole-1) Molecular formula Molecular weight (g mole-1)

IBT 446 CH1.79O0.55N0.15S0.01 24.97 ± 0.01 CH1.85O0.6N0.13S0.008 25.56 ± 0.13

W29 CH1.89O0.56N0.09S0.004 24.36 ± 0.39 CH1.94O0.62N0.1S0.004 25.31 ± 0.32

H222 CH1.9O0.58N0.13S0.005 25.11 ± 0.04 CH1.93O0.64N0.12S0.005 26.04 ± 0.19

327

For transformation of gram dry weight to cmole basis, the average molecular weight of 25.23 g DW cmole -1 was 328

used. 329

Analytical methods 330

For quantifying concentrations of the substrates (glycerol, glucose, xylose and arabinose) and products (xylitol 331

and mannitol) in the culture medium HPLC analysis was performed. The fermentation broth was filtered through 332

a Q-Max® Ca-Plus Filter (Frisenette ApS) with a pore size of 0.45 μm into a HPLC vial which was stored at -20 °C. 333

Separation and detection of the compounds was accomplished with a Bio-Rad Aminex HPX-87H column coupled 334

to a RI detector. Sulphuric acid (5 mM) was used as the mobile phase with a flow velocity of 0.6 ml/min at 60 °C. 335

Data analysis 336

Experiments were performed at least in duplicate (mixed carbon source experiments were performed in 337

triplicate). The maximum specific growth rate (μ max) for all cultivations was estimated through linear regression 338

of OD600 values as a function of time in a semi-logarithmic plot, with a regression correlation of above 0.95. The 339

69

yield coefficients for biomass (Ysx), carbon dioxide (Ysc), xylitol (Ysxy) and mannitol (Ysm) on substrate were 340

estimated using an overall calculation. 341

Acknowledgements 342

PL was supported by a PhD stipend from the Technical University of Denmark. MW acknowledges support from 343

The Danish Council for Strategic Research for Industrial Biotechnology (ERA-IB2). PL performed the experimental 344

work and wrote the manuscript. MW designed the study, supervised the experimental work and co-wrote the 345

manuscript. CTW performed integrated analyses and revision of the manuscript. All authors read and approved 346

the manuscript. The authors declare that they have no competing interests. All data generated or analyzed during 347

this study are included in this published article and its supplementary material files. We acknowledge the 348

Fermentation Platform at Technical University for providing access to fermentation and analytical equipment 349

and for the technical support of Tina Johansen, Alexander Rosenkjaer and Martin Nielsen. We acknowledge the 350

assistance with GC analysis from Silas Anselm Rasmussen. 351

References 352

1. Nicaud J-M. 2012. Yarrowia lipolytica. Yeast 29:409–18. 353

2. Ryu S, Hipp J, Trinh CT. 2015. Activating and Elucidating Metabolism of Complex Sugars in Yarrowia 354

lipolytica. Appl Environ Microbiol 82:1334–45. 355

3. Madzak C, Gaillardin C, Beckerich J-M. 2004. Heterologous protein expression and secretion in the non-356

conventional yeast Yarrowia lipolytica: a review. J Biotechnol 109:63–81. 357

4. Barth G, Gaillardin C. 1997. Physiology and genetics of the dimorphic fungus Yarrowia lipolytica. FEMS 358

Microbiol Rev 19:219–37. 359

5. Beopoulos A, Cescut J, Haddouche R, Uribelarrea J-L, Molina-Jouve C, Nicaud J-M. 2009. Yarrowia 360

lipolytica as a model for bio-oil production. Prog Lipid Res 48:375–87. 361

70

6. Rywińska A, Juszczyk P, Wojtatowicz M, Robak M, Lazar Z, Tomaszewska L, Rymowicz W. 2013. Glycerol 362

as a promising substrate for Yarrowia lipolytica biotechnological applications. Biomass Bioenergy 363

48:148–166. 364

7. Papanikolaou S, Galiotou-Panayotou M, Chevalot I, Komaitis M, Marc I, Aggelis G. 2006. Influence of 365

glucose and saturated free-fatty acid mixtures on citric acid and lipid production by Yarrowia lipolytica. 366

Curr Microbiol 52:134–42. 367

8. Papanikolaou S, Fakas S, Fick M, Chevalot I, Galiotou-Panayotou M, Komaitis M, Marc I, Aggelis G. 2008. 368

Biotechnological valorisation of raw glycerol discharged after bio-diesel (fatty acid methyl esters) 369

manufacturing process: Production of 1,3-propanediol, citric acid and single cell oil. Biomass Bioenergy 370

32:60–71. 371

9. Lazar Z, Dulermo T, Neuvéglise C, Crutz-Le Coq A-M, Nicaud J-M. 2014. Hexokinase—A limiting factor in 372

lipid production from fructose in Yarrowia lipolytica. Metab Eng 26:89–99. 373

10. Westall S, Filtenborg O. 1998. Yeast occurrence in Danish feta cheese. Food Microbiol 15:215–222. 374

11. Workman M, Holt P, Thykaer J. 2013. Comparing cellular performance of Yarrowia lipolytica during 375

growth on glucose and glycerol in submerged cultivations. AMB Express 3:58. 376

12. Klein M, Islam Z-U, Knudsen PB, Carrillo M, Swinnen S, Workman M, Nevoigt E. 2016. The expression of 377

glycerol facilitators from various yeast species improves growth on glycerol of Saccharomyces 378

cerevisiae. Metab Eng Commun 3:252–257. 379

13. Kavšček M, Bhutada G, Madl T, Natter K. 2015. Optimization of lipid production with a genome-scale 380

model of Yarrowia lipolytica. BMC Syst Biol 9:72. 381

14. Mirończuk AM, Biegalska A, Dobrowolski A. 2017. Functional overexpression of genes involved in 382

71

erythritol synthesis in the yeast Yarrowia lipolytica. Biotechnol Biofuels 10:77. 383

15. Ledesma-Amaro R, Lazar Z, Rakicka M, Guo Z, Fouchard F, Coq A-MC-L, Nicaud J-M. 2016. Metabolic 384

engineering of Yarrowia lipolytica to produce chemicals and fuels from xylose. Metab Eng 38:115–124. 385

16. Liang S-H, Wu H, Wang R-R, Wang Q, Shu T, Gao X-D. 2017. The TORC1-Sch9-Rim15 signaling pathway 386

represses yeast-to-hypha transition in response to glycerol availability in the oleaginous yeast Yarrowia 387

lipolytica. Mol Microbiol 104:553–567. 388

17. Pomraning KR, Bredeweg EL, Kerkhoven EJ, Barry K, Haridas S, Hundley H, LaButti K, Lipzen A, Yan M, 389

Magnuson JK, Simmons BA, Grigoriev I V., Nielsen J, Baker SE. 2018. Regulation of Yeast-to-Hyphae 390

Transition in Yarrowia lipolytica. mSphere 3:e00541-18. 391

18. Yuzbasheva EY, Mostova EB, Andreeva NI, Yuzbashev T V., Fedorov AS, Konova IA, Sineoky SP. 2018. A 392

metabolic engineering strategy for producing free fatty acids by the Yarrowia lipolytica yeast based on 393

impairment of glycerol metabolism. Biotechnol Bioeng 115:433–443. 394

19. Mori K, Iwama R, Kobayashi S, Horiuchi H, Fukuda R, Ohta A. 2013. Transcriptional repression by glycerol 395

of genes involved in the assimilation of n-alkanes and fatty acids in yeast Yarrowia lipolytica. FEMS Yeast 396

Res 13:233–40. 397

20. Yang L-B, Zhan X-B, Zheng Z-Y, Wu J-R, Gao M-J, Lin C-C. 2014. A novel osmotic pressure control fed-398

batch fermentation strategy for improvement of erythritol production by Yarrowia lipolytica from 399

glycerol. Bioresour Technol 151:120–7. 400

21. Egermeier M, Russmayer H, Sauer M, Marx H. 2017. Metabolic Flexibility of Yarrowia lipolytica Growing 401

on Glycerol. Front Microbiol 8:49. 402

22. Rywińska A, Rymowicz W, Zarowska B, Skrzypiński A. 2010. Comparison of citric acid production from 403

72

glycerol and glucose by different strains of Yarrowia lipolytica. World J Microbiol Biotechnol 26:1217–404

24. 405

23. Papanikolaou S, Muniglia L, Chevalot I, Aggelis G, Marc I. 2002. Yarrowia lipolytica as a potential 406

producer of citric acid from raw glycerol. J Appl Microbiol 92:737–744. 407

24. Papanikolaou S, Rontou M, Belka A, Athenaki M, Gardeli C, Mallouchos A, Kalantzi O, Koutinas AA, 408

Kookos IK, Zeng A-P, Aggelis G. 2017. Conversion of biodiesel-derived glycerol into biotechnological 409

products of industrial significance by yeast and fungal strains. Eng Life Sci 17:262–281. 410

25. Sherwood KE, Cano DJ, Maupin-Furlow JA. 2009. Glycerol-Mediated Repression of Glucose Metabolism 411

and Glycerol Kinase as the Sole Route of Glycerol Catabolism in the Haloarchaeon Haloferax volcanii. J 412

Bacteriol 191:4307–4315. 413

26. Iida T, Ohta A, Takagi M. 1998. Cloning and characterization of ann-alkane-inducible cytochrome P450 414

gene essential forn-decane assimilation byYarrowia lipolytica. Yeast 14:1387–1397. 415

27. Iida T, Sumita T, Ohta A, Takagi M. 2000. The cytochrome P450ALK multigene family of ann-alkane-416

assimilating yeast,Yarrowia lipolytica: cloning and characterization of genes coding for new CYP52 family 417

members. Yeast 16:1077–1087. 418

28. Ledesma-Amaro R, Lazar Z, Rakicka M, Guo Z, Fouchard F, Coq A-MC-L, Nicaud J-M. 2016. Metabolic 419

engineering of Yarrowia lipolytica to produce chemicals and fuels from xylose. Metab Eng 38:115–124. 420

29. Li H, Alper HS. 2016. Enabling xylose utilization in Yarrowia lipolytica for lipid production. Biotechnol J. 421

30. Rodriguez GM, Hussain MS, Gambill L, Gao D, Yaguchi A, Blenner M. 2016. Engineering xylose utilization 422

in Yarrowia lipolytica by understanding its cryptic xylose pathway. Biotechnol Biofuels 9:149. 423

73

31. Niehus X, Crutz-Le Coq A-M, Sandoval G, Nicaud J-M, Ledesma-Amaro R. 2018. Engineering Yarrowia 424

lipolytica to enhance lipid production from lignocellulosic materials. Biotechnol Biofuels 11:11. 425

32. Tsigie YA, Wang C-Y, Truong C-T, Ju Y-H. 2011. Lipid production from Yarrowia lipolytica Po1g grown in 426

sugarcane bagasse hydrolysate. Bioresour Technol 102:9216–22. 427

33. Spagnuolo M, Shabbir Hussain M, Gambill L, Blenner M. 2018. Alternative Substrate Metabolism in 428

Yarrowia lipolytica. Front Microbiol 9:1077. 429

34. Rodriguez GM, Hussain MS, Gambill L, Gao D, Yaguchi A, Blenner M. 2016. Engineering xylose utilization 430

in Yarrowia lipolytica by understanding its cryptic xylose pathway. Biotechnol Biofuels 9:149. 431

35. Weber C, Farwick A, Benisch F, Brat D, Dietz H, Subtil T, Boles E. 2010. Trends and challenges in the 432

microbial production of lignocellulosic bioalcohol fuels. Appl Microbiol Biotechnol 87:1303–1315. 433

36. Ryu S, Trinh CT, Elliot MA. 2018. Understanding Functional Roles of Native Pentose-Specific Transporters 434

for Activating Dormant Pentose Metabolism in Yarrowia lipolytica. 435

436

437

438

439

440

441

442

74

Supplemental material 443

444

Physiological comparison of Yarrowia lipolytica strains reveals differences in the utilization of sugars and 445

glycerol 446

447

Patrice Lubutaa, Christopher T. Workmana and Mhairi Workmana* 448

aDepartment of Biotechnology and Biomedicine, Technical University of Denmark, Building 223 449

*Present address: Mhairi Workman, Novo Nordisk, Bagsværd, Denmark. 450

451

452

453

454

455

456

457

458

459

460

75

461

Fig. S1 Single carbon cultivations of Y. lipolytica W29 and H222 on glucose. Upper panel: Substrate and product 462

concentrations (cmole L-1) of W29 (A) and H222 (C) cultivations. Bottom panel: Biomass concentration (cmole L-463

1), accumulated CO2 (cmole L-1), and dissolved oxygen level in the broth (%) of W29 (B) and H222 (D) cultivations. 464

The O2 molar ratio of the exhaust gas is also shown. 465

466

467

76

468

Fig. S2 Single carbon cultivations of Y. lipolytica W29 and H222 on glycerol. Upper panel: Substrate and product 469

concentrations (cmole L-1) of W29 (A) and H222 (C) cultivations. Bottom panel: Biomass concentration (cmole L-470

1), accumulated CO2 (cmole L-1), and dissolved oxygen level in the broth (%) of W29 (B) and H222 (D) cultivations. 471

The O2 molar ratio of the exhaust gas is also shown. 472

473

474

77

475

Fig. S3 Full length mixed carbon cultivation of Y. lipolytica IBT 446. (A) Substrate and product concentrations 476

(cmole L-1). (B) Biomass concentration (cmole L-1), accumulated CO2 (cmole L-1), and dissolved oxygen level in the 477

broth (%). The O2 molar ratio of the exhaust gas is also shown. 478

479

480

481

482

78

483

Fig. S4 Full length mixed carbon cultivation of Y. lipolytica W29. (A) Substrate and product concentrations (cmole 484

L-1). (B) Biomass concentration (cmole L-1), accumulated CO2 (cmole L-1), and dissolved oxygen level in the broth 485

(%). The O2 molar ratio of the exhaust gas is also shown. 486

487

488

489

79

490

Fig. S5 Full length mixed carbon cultivation of Y. lipolytica H222. (A) Substrate and product concentrations (cmole 491

L-1). (B) Biomass concentration (cmole L-1), accumulated CO2 (cmole L-1), and dissolved oxygen level in the broth 492

(%). The O2 molar ratio of the exhaust gas is also shown. 493

80

Manuscript 2: Draft Genome Sequences of Yarrowia 1

lipolytica Strains H222, IBT 446 and W29 2

3

Patrice Lubutaa, Mhairi Workmana and Christopher T. Workmana# 4

aDepartment of Biotechnology and Biomedicine, Technical University of Denmark, Building 223, Søltofts Plads, 5

2800 Kgs. Lyngby, Denmark 6

#Address correspondence to Christopher T. Workman, cwor@dtu.dk. 7

Running title: Y. lipolytica IBT446, H222 and W29 draft genomes 8

9

10

11

12

13

14

15

16

17

18

81

Abstract 19

Yarrowia lipolytica is a non-conventional yeast with a high potential for various biotechnological applications 20

and is also used as a model organism in cell biological research. Here we report the draft genome sequence of 21

two new strains H222 and IBT 446 and a re-sequenced draft genome of strain W29. 22

Introduction 23

The number of sequenced Y. lipolytica strains has increased since the first genome was available in 2004, and 24

today two high-quality reference genomes are available for CLIB 122 and W29 strains (1, 2). Additionally, draft 25

genomes are available for PO1f, W29, WSH-Z06 and A101 strains (3–5). Here we report the availability of new 26

draft genome sequences for Y. lipolytica H222 and IBT 446 as well as a re-sequencing of W29. Y. lipolytica H222 27

has been used in studies addressing sucrose conversion and hydrophobic substrate utilization, while is a 28

Technical University of Denmark strain originally isolated from feta cheese (6, 7). IBT 446 shows unique 29

properties regarding substrate utilization and yeast-to-hyphae transition (Lubuta et al. 2018, submitted for 30

publication). We further re-sequenced the W29 strain used in our physiology studies to confirm the genotype of 31

the reference. The increasing availability of Y. lipolytica genomes will facilitate insight into the genetic variation 32

of this important yeast. 33

Results and Discussion 34

Genomic DNA of the three strains was sequenced with an Illumina MiSeq instrument in paired-end mode 35

generating 9.3, 8.6 and 8.3 million reads for IBT 446, H222 and W29 respectively. Raw reads were quality checked 36

with the FastQC tool version 0.11.5 (8) and subsequently adapter and quality trimmed with Trimmomatic version 37

0.36 (9) and BBDuk version 37.95 (10) . Quality trimming resulted in 7.69M read-pairs for IBT 446, 7.03M read-38

pairs for H222, and 6.78M read-pairs for W29. Coverage was estimated from the sum of all nucleotides in the 39

trimmed reads relative to the size of YALI1 genome (20.84 Mbps) and was found to be 92X for IBT 446, 82X for 40

H222, and 73X for W29. Spades version 3.11.1 was used for de novo assembly of the trimmed reads (11). The 41

82

assembly resulted in 4157 contigs (19975838 bases; N50 = 8196; 3911 contigs > 500 bp) for IBT 446, 4292 contigs 42

(19902020 bases; N50 = 7831; 4014 contigs > 500 bp) for H222 and 4457 contigs (19922323 bases; N50 = 7666; 43

4150 contigs > 500 bp) for W29. Finally, the reference-based genome arrangement tool Chromosomer (version 44

0.1.3) (12) was used to build chromosomes by aligning contigs to the existing reference genome of W29 (YALI1, 45

GenBank accession: GCA_001761485.1). The draft genome assemblies consist of six nuclear chromosomes with 46

100 Ns representing gaps. 47

Mauve version 2.4.0 (13) was used to align each chromosome of W29, H222 and IBT 446 to those of CLIB122 and 48

YALI1 reference genomes, giving a multiple alignment of 5 strains for each nuclear chromosome. This alignment 49

identified 72801 single-nucleotide variants (SNV), similar to SNPs, which was used to analyze differences 50

between the 5 strains. Our W29 genome only differed in 6% of the 72801 SNVs from YALI1, indicating a maximum 51

error rate in our analysis due to errors in alignment or sequence assembly. Both H222 and IBT 446 differed 52

significantly from W29, YALI1 and CLIB122, where difference rates were observed in the range of 56 to 62% of 53

all SNVs. This was in contrast to the degree of similarity found between H222 and IBT 446 that differed in less 54

than 21% of all SNVs. 55

Accession number(s). Chromosomes of the non-Whole Genome Shotgun assemblies have been deposited in 56

GenBank under the accession no. CP028454.1-CP028459.1 (IBT 446), CP028442.1-CP028447.1 (H222) and 57

CP028448.1-CP028453.1 (W29). All versions described in this paper are the versions 1.0. 58

Acknowledgments 59

This study was funded by the ERA-NET scheme of the 7th EU Framework Program Integrated Process and Cell 60

Refactoring Systems for Enhanced Industrial Biotechnology (IPCRES). 61

62

83

References 63

1. Dujon B, Sherman D, Fischer G, Durrens P, Casaregola S, Lafontaine I, De Montigny J, Marck C, Neuvéglise 64

C, Talla E, Goffard N, Frangeul L, Aigle M, Anthouard V, Babour A, Barbe V, Barnay S, Blanchin S, Beckerich 65

J-M, Beyne E, Bleykasten C, Boisramé A, Boyer J, Cattolico L, Confanioleri F, De Daruvar A, Despons L, 66

Fabre E, Fairhead C, Ferry-Dumazet H, Groppi A, Hantraye F, Hennequin C, Jauniaux N, Joyet P, Kachouri 67

R, Kerrest A, Koszul R, Lemaire M, Lesur I, Ma L, Muller H, Nicaud J-M, Nikolski M, Oztas S, Ozier-68

Kalogeropoulos O, Pellenz S, Potier S, Richard G-F, Straub M-L, Suleau A, Swennen D, Tekaia F, 69

Wésolowski-Louvel M, Westhof E, Wirth B, Zeniou-Meyer M, Zivanovic I, Bolotin-Fukuhara M, Thierry A, 70

Bouchier C, Caudron B, Scarpelli C, Gaillardin C, Weissenbach J, Wincker P, Souciet J-L. 2004. Genome 71

evolution in yeasts. Nature 430:35–44. 72

2. Magnan C, Yu J, Chang I, Jahn E, Kanomata Y, Wu J, Zeller M, Oakes M, Baldi P, Sandmeyer S. 2016. 73

Sequence Assembly of Yarrowia lipolytica Strain W29/CLIB89 Shows Transposable Element Diversity. PLoS 74

One 11:e0162363. 75

3. Liu L, Alper HS. 2014. Draft Genome Sequence of the Oleaginous Yeast Yarrowia lipolytica PO1f, a 76

Commonly Used Metabolic Engineering Host. Genome Announc 2. 77

4. Pomraning KR, Baker SE. 2015. Draft Genome Sequence of the Dimorphic Yeast Yarrowia lipolytica Strain 78

W29. Genome Announc 3. 79

5. Devillers H, Brunel F, Połomska X, Sarilar V, Lazar Z, Robak M, Neuvéglise C. 2016. Draft Genome Sequence 80

of Yarrowia lipolytica Strain A-101 Isolated from Polluted Soil in Poland. Genome Announc 4. 81

6. Westall S, Filtenborg O. 1998. Yeast occurrence in Danish feta cheese. Food Microbiol 15:215–222. 82

7. Workman M, Holt P, Thykaer J. 2013. Comparing cellular performance of Yarrowia lipolytica during growth 83

84

on glucose and glycerol in submerged cultivations. AMB Express 3:58. 84

8. Andrews S. 2010. FastQC: a quality control tool for high throughput sequence data. Available online at: 85

http://www.bioinformatics.babraham.ac.uk/projects/fastqc. 86

9. Bolger AM, Lohse M, Usadel B. 2014. Trimmomatic: a flexible trimmer for Illumina sequence data. 87

Bioinformatics 30:2114–20. 88

10. Brian Bushnell. 2014. BBtools package. Dep Energy Jt Genome Inst (DOE JGI). 89

11. Bankevich A, Nurk S, Antipov D, Gurevich AA, Dvorkin M, Kulikov AS, Lesin VM, Nikolenko SI, Pham S, 90

Prjibelski AD, Pyshkin A V, Sirotkin A V, Vyahhi N, Tesler G, Alekseyev MA, Pevzner PA. 2012. SPAdes: a 91

new genome assembly algorithm and its applications to single-cell sequencing. J Comput Biol 19:455–77. 92

12. Tamazian G, Dobrynin P, Krasheninnikova K, Komissarov A, Koepfli K-P, O’Brien SJ. 2016. Chromosomer: 93

a reference-based genome arrangement tool for producing draft chromosome sequences. Gigascience 94

5:38. 95

13. Darling ACE, Mau B, Blattner FR, Perna NT. 2004. Mauve: multiple alignment of conserved genomic 96

sequence with rearrangements. Genome Res 14:1394–403. 97

98

85

Manuscript 3: Genome-wide expression analysis of 1

Yarrowia lipolytica strains varying in the utilization 2

of glucose and glycerol 3

4

Patrice Lubutaa, Mhairi Workmana*, Eduard Kerkhoven#b & Christopher T. Workman#a 5

aDepartment of Biotechnology and Biomedicine, Technical University of Denmark, Kgs. Lyngby, Denmark 6

bDepartment of Biology and Biological Engineering, Systems and Synthetic Biology, Chalmers University of 7

Technology, Gothenburg, Sweden 8

9

Running Head: Transcriptome analysis of Y. lipolytica strains 10

11

12

#Address correspondence to Christopher T. Workman, cwor@dtu.dk. 13

#Address correspondence to Eduard Kerkhoven, eduardk@chalmers.se. 14

*Present address: Mhairi Workman, Novo Nordisk, Bagsværd, Denmark. 15

16

Keywords: Yarrowia lipolytica, quantitative physiology, transcriptomics, glycerol, glucose, carbon 17

repression 18

86

Abstract 19

Glycerol is considered as a promising substrate for biotechnological applications and the non-conventional yeast 20

Yarrowia lipolytica has been used extensively for the valorization of this compound. Contrary to S. cerevisiae, 21

Y. lipolytica seems to prefer glycerol over glucose and it has been reported previously that the presence of 22

glycerol can suppress the consumption of glucose during co-substrate cultivations. Additionally, it has been 23

shown that genes related to n-alkane utilization are transcriptionally repressed by glycerol. Based on these 24

observations, we hypothesized glycerol repression-like effects in Y. lipolytica, which are converse to well 25

described carbon repression mechanisms ensuring the prioritized use of glucose. We therefore aimed to 26

investigate this effect on the level of gene expression. Strains varying in the degree of glucose suppression were 27

chosen based on previous results, and analyzed in high-resolution growth screenings, resulting in the detection 28

of different growth phenotypes. The strains IBT 446 and W29 were selected for chemostat cultivations on 29

glucose, glycerol and mixed carbon conditions, followed by an RNAseq-based transcriptome analysis. We could 30

show that several transporters were significantly higher expressed in W29, however, the major differences in 31

expression between the strains were regardless of the carbon source applied. Cross-comparisons revealed that 32

the strain-specific carbon responses went in the opposite direction. Finally, further analysis led to the 33

identification of several differentially expressed genes related to transcriptional regulation and signal 34

transduction. This study provides an initial investigation and paves the way for future investigations on 35

potentially novel carbon source regulation mechanisms in the non-conventional yeast Y. lipolytica. 36

37

38

87

Introduction 39

Glycerol, a by-product from the biodiesel production is considered as a promising substrate for biotechnological 40

applications. The biodiesel industry increased rapidly in the European Union and the U.S. over the last fifteen 41

years, leading to an increased availability of crude glycerol and a drastic decrease of its market price (Valerio et 42

al. 2015). The use of glycerol by microbial fermentation makes high efficient production strains (so-called cell 43

factories) indispensable. Saccharomyces cerevisiae is among yeasts the most established microorganism applied, 44

however, its natural capacity to utilize this substrate is limited (Klein et al. 2017). In contrast, several other yeast 45

species are naturally superior glycerol users, for instance, Pachysolen tannophilus, Pichia pastoris, 46

Cyberlindnera jadinii or Yarrowia lipolytica (Klein et al. 2016). The oleaginous yeast Y. lipolytica gathered 47

attention in recent years, especially due to its ability to produce economically interesting compounds (Liu, Ji, and 48

Huang 2015). Growth rates of Y. lipolytica on glycerol exceeds levels of 0.4 h-1 (Klein et al. 2016) and various 49

attempts aimed to convert glycerol into value-added products (Rywińska et al. 2013). 50

Glycerol metabolism has been most studied in S. cerevisiae (Klein et al. 2017). Both species, S. cerevisiae and 51

Y. lipolytica are using the glycerol-3-phosphate pathway in order to metabolize glycerol (Dulermo and Nicaud 52

2011; Sprague and Cronan 1977; Pavlik et al. 1993), however, several differences in glycerol uptake, the presence 53

of metabolic enzymes and carbon regulation exist. In contrast to S. cerevisiae, Y. lipolytica seems to prefer 54

glycerol over glucose as a source of carbon and energy. It could be shown that in single carbon cultivations growth 55

rates on glycerol are higher than those on glucose, and additionally, that the consumption of glucose is 56

suppressed in glucose-glycerol co-cultivations (Workman, Holt, and Thykaer 2013; Mori et al. 2013; Yuzbasheva 57

et al. 2018). Interestingly, the glucose consumption restores after the depletion of glycerol. These observations 58

point to carbon regulation mechanisms allowing Y. lipolytica the prioritized use of glycerol. The underlying 59

mechanisms have not been elucidated yet but must differ drastically from well-known carbon catabolite 60

repression (CCR) mechanisms (e.g. in S. cerevisiae or E. coli) ensuring the prioritized use of glucose (Gancedo 61

88

1998; Brückner and Titgemeyer 2006). For instance, S. cerevisiae genes related to glycerol uptake (STL1) and 62

catabolism (GUT1, GUT2) are repressed under glucose and derepressed after its depletion when growth occurred 63

on non-fermentable carbon sources (Morten Grauslund, Lopes, and Rønnow 1999; M Grauslund and Rønnow 64

2000; Ferreira et al. 2005). 65

This study provides an initial investigation on potentially novel carbon source regulation mechanism in the non-66

conventional yeast Y. lipolytica. Known carbon regulatory mechanisms act on the level of transcription, and 67

therefore, our approach aimed to investigate Y. lipolytica´s transcriptome. So far, glycerol mediated repression 68

of glucose utilization has only been described for the haloarchaeon Haloferax volcanii (Sherwood et al. 2009). 69

However, it could be shown that n-alkane utilization of Y. lipolytica is transcriptionally repressed by glycerol (Iida 70

et al. 2000; Iida, Ohta, and Takagi 1998; Mori et al. 2013). Interestingly, the above mentioned glycerol induced 71

suppression of glucose consumption in co-substrate cultivations seems to be strain dependent. While most 72

strains showed glycerol repression-like effects, some strains were able to use glycerol and glucose 73

simultaneously (Lubuta, et al (2018), manuscript under review). We therefore tried to gain insights from the 74

analysis of these strains: In initial experiments, the growth physiology was investigated by high-frequent biomass 75

measurements in order to identify diauxic shift-like events during mixed substrate cultivations. The strains IBT 76

446 and W29 were selected and grown in chemostats using glycerol, glucose and a glycerol-glucose blend as 77

carbon sources. Samples were taken and analyzed by RNA-seq based transcriptomics in order to compare the 78

transcriptional profiles. 79

80

81

82

89

Materials and Methods 83

Yeast strains and media 84

Three wild type Y. lipolytica strains were used throughout this study: Y. lipolytica W29 (CLIB 89) and Y. lipolytica 85

H222 (CLIB 80) were obtained from CIRM-Levures strain collection, Institute National de la Recherche 86

Agronomique (INRA, France). Y. lipolytica IBT 446 was obtained from the culture collection of the Department of 87

Biotechnology and Biomedicine, Technical University of Denmark (DTU). For long-term storage, strains were 88

grown in YPD liquid media (1% yeast extract, 2% glucose, 2% peptone) and kept at -80 °C in 17 % (v/v) glycerol. 89

YPD plates were used for short-term storage and the strains were grown for 2 days at 30 °C. YPD plates were 90

stored at 4 °C. All cultivation experiments were performed in defined minimal media as described in (Workman, 91

Holt, and Thykaer 2013). 92

Microscale cultivations 93

A micro-scale fermentation system (BioLector, m2p-Labs GmbH) was used to screen for growth differences when 94

varying glycerol-glucose blends were used as carbon sources. Cultivations took place in 48-well microtiter plates 95

(MTP-48-B Flowerplates, m2p-Labs GmbH) with a working volume of 1.5 ml and 1000 rpm shaking speed. The 96

temperature was maintained at 30 °C, and humidity control was active to reduce evaporation. Online monitoring 97

of biomass accumulation was achieved by light scattering measurement at 620 nm approximately every 3 98

minutes. Table 1 shows the used glycerol and glucose concentrations. Precultures were grown in shake flasks 99

using defined minimal media and 20 g L-1 glycerol as the carbon source. Cells were harvested in mid exponential 100

phase, and washed to remove residual substrate. Experiments were conducted with at least 4 replicates and in 101

independent plate runs. 102

103

104

90

Table 1. Glycerol and glucose concentrations used in the growth experiments. 105

# Glycerol : Glucose Ratio Glycerol concentration [mol L-1] Glucose concentration [mol L-1]

1. 1 : 0 0.054 0

2. 2 : 1 0.036 0.018

3. 1 : 1 0.027 0.027

4. 1 : 2 0.018 0.036

5. 1 : 4 0.011 0.043

6. 0 : 1 0 0.054

106

Chemostat cultivations 107

Chemostat cultivations were carried out in order to generate biomass samples used for mRNA extraction. 108

Cultivations were conducted in fully instrumented and automatically controlled 1 L BIOSTAT® Q plus fermenters 109

(Sartorius Stedim Biotech S.A) with a working volume of 0.5 L. Cells were grown in batch mode until late 110

exponential phase (determined by CO2 exhaust gas measurement) before to the continuous mode was initiated. 111

Liquid in and out flows were controlled gravimetrically. Carbon limited conditions were applied and the dilution 112

rate was adjusted to D = 0.1 h-1. Three experimental conditions have been tested: glycerol 10 g L-1 (≈ 0.11 mole 113

L-1), glucose 10 g L-1 (≈ 0.06 mole L-1) and a mix of glycerol 5 g L-1 (≈ 0.05 mole L-1) and glucose 5 g L-1 (≈ 0.03 mole 114

L-1). All cultivations were applied in triplicates resulting in 18 total chemostat cultivations. 115

The biomass concentration was determined by cell dry weight estimation using 0.45 μm nitrocellulose filters 116

(Sartorius Stedium) for broth filtration and microwave desiccation (150 W for 20 min). HPLC analysis was used 117

to quantify substrate concentrations. The fermentation broth was filtered and compounds were separated by an 118

Aminex HPX-87H column (Bio-Rad) prior the detection via RI detector. Off gas analysis was carried out by mass 119

spectrometry using a Prima PRO Process Mass Spectrometer (Thermo Fisher Scientific) quantifying the exhaust 120

gas composition. Biomass samples were taken under steady state conditions. The broth was centrifuged in 2 ml 121

aliquots and cell pellets immediately frozen in liquid nitrogen. Biomass samples were kept at -80 °C until further 122

use. 123

91

RNA extraction and sequencing 124

Cell pellets were disrupted and homogenized by bead-milling in a TissueLyser (Quiagen) and the use of metal 125

beads. RNA extraction was carried with the RNeasy® Plus Mini Kit (cat. nos. 74134, Quiagen) according the 126

standard protocol. Samples were barcoded, multiplexed and sequenced using a HiSeq 4000 instrument (illumina) 127

in paired end mode. Reads with a length of 150 base pairs were generated. 128

Transcriptome data analysis 129

Raw reads were demultiplexed with the Barcode Splitter tool from the FASTX toolkit version 0.0.14 (Hannon lab). 130

The raw reads were subsequently quality controlled with the FastQC tool version 0.11.5 (Andrews 2010) and 131

quality trimmed with Trimmomatic version 0.36 (Bolger, Lohse, and Usadel 2014). Read mapping and 132

quantification was carried out with the Subread package (Liao, Smyth, and Shi 2013) using the W29 genome as 133

a reference (GenBank assembly accession: GCA_001761485.1) (Magnan et al. 2016). In order to facilitate the 134

comparability between W29 gene identifiers (YALI1_ID) and the older CLIB 122 (GenBank assembly accession: 135

GCA_000002525.1) identifiers (YALI0_ID), we provide both identifiers in every table. 136

Raw read counts have been converted into transcripts per million (TPM) according (Wagner, Kin, and Lynch 137

2012), to compare the expression of different genes across the samples. A differential gene expression analysis 138

was carried out using the edgeR package (Robinson, McCarthy, and Smyth 2010) for importing, filtering and 139

normalizing raw count data and the limma package for linear modelling (Law et al. 2014). 140

Several linear models have been used throughout the study: In order to extract the strain effect we used a model 141

describing the expression as function of strain effect (s) and carbon source condition effect (c): 𝒚 = 𝒔𝒙 + 𝒄𝒙 +142

𝝐. The strain term was categorical while the condition term was assumed to be ordinal resulting in a linear 143

coefficient and a quadratic coefficient. Further, to analyze strain-specific responses to the applied carbon 144

sources, cross-comparisons between samples have been carried out. Therefore, the strain and condition factors 145

92

were combined into one factor (e.g. IBT.Glycerol) and comparisons of interest were extracted as contrasts. 146

Finally, to investigate the influence of the different carbon sources across the two strains, we formulated three 147

models with separate factors for glucose (cglu) and glycerol (cgly) (present vs not-present). In order to extract 148

genes responding to the presence of glucose in both strains we formulated model 1: 𝒚 = 𝒔𝒙 + 𝒄𝒈𝒍𝒖𝒙 + 𝝐. To 149

extract genes responding to the presence of glycerol in both strains we formulated model 2: 𝒚 = 𝒔𝒙 + 𝒄𝒈𝒍𝒚𝒙 +150

𝝐. Finally, to extract genes differently responding in the two strains we formulated model 3: = 𝒔𝒙 + 𝒄𝒈𝒍𝒚,𝑰𝑩𝑻𝒙 +151

𝒄𝒈𝒍𝒖,𝑾𝟐𝟗𝒙 + 𝝐 , where we specifically modeled the factor glycerol and IBT 446 , and the factor glucose and W29. 152

Gene set analysis 153

GO term annotations of the Y. lipolytica W29 genome (biosample: SAMN04088558) were assigned by Blast2GO 154

(Conesa et al. 2005) using the provided fungi reference database and InterProScan (Jones et al. 2014) using 155

default settings. The Piano R-package was used for gene-set analyses (GSA) (Väremo, Nielsen, and Nookaew 156

2013). Only gene sets with more than 5 and less than 500 genes were included. Piano´s consensus gene-set 157

analysis function was used to condense results from several different GSA methods. 158

159

160

161

93

Results and Discussion 162

Glycerol-glucose mixed cultivations revealed physiological differences between 163

Y. lipolytica W29 and IBT 446 164

Our previous results showed that the Y. lipolytica strain IBT 446 exhibits a sequential substrate utilization of 165

glycerol and glucose, while the strains W29 and H222 exhibit a higher degree of co-consumption (Lubuta et al. 166

2018, manuscript under review). Based on these observations, we supposed carbon repression-like mechanisms 167

ensuring the prioritized use of glycerol and that these mechanism are additionally strain dependent in 168

Y. lipolytica. In a first attempt to investigate these phenomena, it should be determined if these strain dependent 169

substrate utilization phenotypes have an effect on growth when glycerol-glucose mixtures are applied. 170

Microscale cultivations with high frequent biomass measurements (approximately every 3 min) were used to 171

detect small changes in the biomass accumulation. Since Y. lipolytica grows faster on glycerol (µ = 0.30 h-1) than 172

on glucose (µ = 0.24 h-1) (Workman 2013), two growth phases should be visible for strains exhibiting a sequential 173

consumption, whereas only one growth phase should be present if strains exhibiting co-consumption. 174

Additionally, a short second lag phase between the consumption of glycerol and glucose was observed by strains 175

with sequential uptake (Lubuta et al. 2018, manuscript under review). The three Y. lipolytica strains W29, H222 176

and IBT 446 were tested on six different glycerol-glucose ratios and growth profiles are shown in Figure 1. 177

178

94

179

Figure 1. Growth profiling by micro scale cultivations using different ratios of glycerol and glucose (see Table 1). (A,B,C): 180

Growth profiles of W29, H222 and IBT 446. (D) Correlation between the glycerol molar fraction in the media and glycerol 181

attenuation fraction in growth experiments with IBT 446. 182

All three strains showed an initial lag phase, which was longer when glucose was the only carbon source. This 183

was also the case when precultures were grown on glucose instead of glycerol (data not shown). After the initial 184

lag phase all strains grew exponentially. The high-resolution growth profiles revealed two types of transitions in 185

the biomass accumulation: All growth curves of W29 and H222 showed a modest increase in the growth half-186

way through their cultivation time (20-25 h, indicated by black arrows in Figure 1A and B). However, this 187

transition was observable under all conditions (including the 1:0 and 0:1 ratios), and therefore, a specific 188

95

response to the varying ratios was excluded. We assumed this transition indicates morphological changes, since 189

hyphae formation was detected by microscopy (data not shown). Interestingly, another type of transition was 190

observable in cultivations with IBT 446: Here, two distinct growth phases were distinguishable, whereby the first 191

one increased in its length the more glycerol was available (dashed line in Figure 1C). A significant correlation 192

was observed between the proportion of biomass generated before an observable diauxic shift (the glycerol 193

attenuation fraction) and the molar fraction of glycerol in the media (Figure 1D). In contrast, it was not possible 194

to link the substrate molar fractions to the transition phases in W29 and H222 cultivations. These results support 195

a sequential substrate utilization by IBT 446 which has a direct effect on the growth. Based on these findings we 196

formulated the hypothesis, that genes related to glucose utilization are subject to a glycerol induced repression 197

in IBT 446, while this effect is absent or reduced in W29 and H222 (Figure 2). To test this hypothesis, we selected 198

the strains IBT 446 and W29 for chemostat cultivations and a subsequent transcriptome analysis. The gene 199

expression data was used to investigate if observed physiological differences were linked to differences in gene 200

expression. 201

202

203

Figure 2. Hypothesis for the explanation of observed phenotypical differences between IBT 446 and W29. The glucose and 204

glycerol catabolic routes are connected over the common metabolite DHAP. We hypothesized glycerol repression-like 205

effects in IBT 446 preventing the simultaneous consumption of glucose in the presence of glycerol. Potential targets of this 206

repression are glucose uptake or genes of the upper glycolysis (both in red). GAP: glyceraldehyde-3-phosphate. DHAP: 207

dihydroxyacetonephosphate. 208

96

Chemostat cultivations revealed lower respiration rates in IBT 446 209

Chemostat cultivations with the strains Y. lipolytica IBT 446 and W29 were conducted in order to gain biomass 210

samples for a subsequent transcriptome analysis. Three different conditions were applied: defined minimal 211

media with either glycerol, glucose or a glycerol-glucose mix. The chemostats were carbon limited with a dilution 212

rate adjusted to 0.1 h-1. Transcriptomes provided during growth on single carbon sources were then compared 213

with the glycerol-glucose mixed condition revealing potential differences in carbon source regulation between 214

the two strains. Table 2 shows the main physiological parameters of the chemostat experiments. Due to carbon 215

limited conditions, substrate concentrations in the bioreactor were not detected throughout all conditions (0 g 216

L-1). Specific substrate consumption rates in mmole substrate gDW-1 h-1 were roughly twice those for glycerol 217

compared to glucose, since glycerol has only half of the molecular weight of glucose (92.09 g mol−1 vs 180.16 g 218

mol−1) resulting in the double amount of moles used in the cultivations. Since the biomass concentration of W29 219

under mixed conditions was slightly lower than in the other cultivations, calculations led to slightly higher specific 220

substrate consumption rates (qGlu and qGly). Specific oxygen consumption rates qO2 and carbon dioxide production 221

rates qCO2 of both strains varied throughout the applied conditions. In glucose cultivations oxygen consumption 222

and carbon dioxide production had nearly the same values which is reflected by a respiratory quotient (RQ) of 223

close to one. In contrast, oxygen consumption was higher than the carbon dioxide production when grown under 224

glycerol, giving a RQ of 0.67 (IBT 446) and 0.69 (W29). In mixed substrate cultivations oxygen consumption rates 225

and carbon dioxide production rates showed values in between the single carbon cultivations, resulting in RQ 226

values of 0.84 (IBT 446) and 0.85 (W29). Interestingly, qO2 and qCO2 were generally higher in the W29 cultivations 227

throughout all conditions, which could indicate that W29 has a more active oxidative phosphorylation. However, 228

the biomass yields Ysx for IBT 446 and W29 were similar throughout almost all conditions: roughly 65 % of the 229

carbon went into biomass (cmole cmole-1). For both strains carbon dioxide yields Ysc were higher on glucose 230

compared to the other conditions. Only carbon yields from W29 cultivations under glucose add up to one. In the 231

97

other cultivations some carbon was unaccounted (approximately 10 %), indicating the secretion of undetected 232

by-products. 233

234

Table 2. Physiological parameters of the carbon limited chemostat experiments at steady state. The strains IBT 446 and W29 235

have been cultivated on glucose, glycerol and a glucose-glycerol mix with a dilution rate of 0.1 h-1. RQ: respiratory quotient. 236

DO: dissolved oxygen. 237

IBT 446 glucose

IBT 446 glycerol

IBT 446 glucose/glycerol

W29 glucose

W29 glycerol

W29 glucose/glycerol

Biomass conc. (g L-1) 5.3 ± 0.2 5.3 ± 0.3 5.4 ± 0.4 5.1 ± 0.2 5.3 ± 0.0 4.6 ± 0.1

qGlu (mmole gDW-1 h-1) -1.04 ± 0.03 0.00 ± 0.00 -0.51 ± 0.04 -1.08 ± 0.04 0.00 ± 0.00 -0.61 ± 0.02

qGly (mmole gDW-1 h-1) 0.00 ± 0 -2.04 ± 0.11 -1.01 ± 0.07 0.00 ± 0.00 -2.05 ± 0.01 -1.19 ± 0.03

qO2 (mmole gDW-1 h-1) -1.67 ± 0.05 -2.17 ± 0.11 -1.75 ± 0.28 -2.23 ± 0.06 -2.49 ± 0.07 -2.39 ± 0.12

qCO2 (mmole gDW-1 h-1) 1.81 ± 0.02 1.46 ± 0.06 1.46 ± 0.22 2.41 ± 0.02 1.71 ± 0.06 2.03 ± 0.07

Ysx (cmole cmole-1) 0.64 ± 0.02 0.66 ± 0.04 0.66 ± 0.04 0.62 ± 0.02 0.65 ± 0.0 0.55 ± 0.02

Ysc (cmole cmole-1) 0.29 ± 0.01 0.24 ± 0.02 0.24 ± 0.03 0.37 ± 0.02 0.28 ± 0.01 0.28 ± 0.01

RQ (-) 1.08 ± 0.04 0.67 ± 0.03 0.84 ± 0.02 1.08 ± 0.02 0.69 ± 0.01 0.85 ± 0.01

DO (%) 66 ± 12 45 ± 3 59 ± 2 51 ± 3 40 ± 1 55 ± 2

Carbon source concentration for all substrates at steady state: 0 g L-1. 238

239

240

241

98

Evidence for differences in physiology of carbon utilization are observed in mRNA levels of 242

targeted genes 243

RNA samples obtained from chemostat cultivations were sequenced and resulting reads have been quantified 244

using the W29 genome (GenBank assembly accession: GCA_001761485.1) as a reference. The genome-wide 245

expression data was analyzed by two approaches: in the targeted analysis genes directly involved in glycerol and 246

glucose metabolism and transport have been investigated, while in the explorative approach linear modeling 247

was used to systematically analyze the effects caused by the experimental factors. YALI1 gene identifiers 248

have been used throughout this study, but YALI0 identifiers are provided in tables to facilitate comparability 249

with the older CLIB 122 YALI0 identifiers (GenBank assembly accession: GCA_000002525.1). 250

Significant strain differences were observed between glucose and glycerol transporters: Only a few studies 251

addressed glucose and glycerol uptake in Y. lipolytica. One attempt to decipher sugar transport mechanisms in 252

Y. lipolytica resulted in the identification of 24 proteins related to hexose transport (Lazar et al. 2017). The 253

authors showed that these putative sugar porters are distributed among six different clusters (class A to F) in a 254

phylogenetic analysis. Furthermore, six out of the 24 proteins were identified to be hexose transporters (named 255

Yarrowia Hexose Transporter: YHT1 to YHT6) and among them YHT1 and YHT4 seem to be most important for 256

hexose uptake. In the present study, we used the nomenclature presented by (Lazar et al. 2017) and investigated 257

the identified putative transporters. There is evidence that glycerol uptake in Y. lipolytica (and in other glycerol 258

utilizing yeasts: e.g. P. tannophilus, K. pastoris, and C. jadinii) is mediated by a homolog to S. cerevisiae 259

aquaglyceroporin FPS1 (Klein et al. 2016). This is in contrast to S. cerevisiae where glycerol uptake is solely 260

mediated by the glycerol/H+ symporter Stl1 (Ferreira et al. 2005). Therefore, we decided to investigate the 261

expression levels of all genes putatively related to glycerol and sugar uptake in Y. lipolytica (Table S1), to get 262

insights if differences in transporter expression contribute to the observed physiological differences. 263

99

Raw count values (Table S2) were converted to transcripts per million (TPM) to allow for comparison of genes 264

across samples (Table S3). The expression levels of genes related to glycerol and sugar transport are provided in 265

Figure 3A. Based on their level of expression, the FPS1 homolog YlFPS1 (YALI1_F00616g), YHT1 (YALI1_C08523g) 266

and YHT4 (YALI1_E27441g) are the dominating transport related genes under the applied conditions. 267

Interestingly, levels of YlFPS1 and YHT1 transcripts were significantly higher in W29 than in IBT 446. In both 268

strains, expression of YlFPS1 was strongly induced by glycerol, evidencing a transcriptionally-regulated role of 269

this transporter in the assimilation of glycerol. Contrary, the expression of YHT1 was not much affected by a 270

specific conditions in W29, whereas in IBT 446 glycerol had a minor positive effect on its expression. YHT4 was 271

slightly higher expressed in IBT and also upregulated in the presence of glycerol, while in W29 this gene is majorly 272

upregulated under glucose. Furthermore, two transporters YALI1_D00376g (class D) and YALI1_F24031g (class 273

C) were nearly exclusively expressed in W29. For various putative transporter genes, expression levels were very 274

low or absent in any of the tested conditions. 275

As stated above, we hypothesized that genes related to glucose transport or catabolism are subject to a glycerol 276

induced repression in IBT 446. In our targeted study, however, we did not observe a repression on genes related 277

to hexose transport. Nevertheless, the three transporters YALI1_C08523g (YHT1), YALI1_D00376g and 278

YALI1_F24031g were significantly higher expressed in W29 throughout all conditions. This observation could 279

potentially be related to the absence of glycerol attenuation in W29. Interestingly, YHT1 is closely related to the 280

glucose sensors SNF3 and RGT3 in S. cerevisiae (Lazar et al. 2017), while this would have to be further 281

investigated to elucidate a potential relationship between these transporters and the observed physiological 282

effects. 283

284

285

100

286

Figure 3. Results of the targeted transcriptome analysis. Expression levels are shown in log transcripts per million (logTPM) 287

and names of S. cerevisiae orthologs are provided. (A): Expression of genes related to glycerol and sugar transport (Table S1 288

for gene information). (B): Expression levels of genes related to glycerol metabolism (Table S4 for gene information). 289

101

No evidence for glycerol repression observed in mRNA levels of glycolytic genes: Besides genes related to 290

glucose transport, we hypothesized that genes involved in glucose catabolism could be other potential targets 291

for a repression by glycerol. The catabolic routes of glycerol and glucose are connected via the intermediate 292

DHAP. Therefore, we speculated that genes encoding enzymes of the upper glycolysis (before DHAP) could be 293

repressed in IBT 446 but not in W29. However, transcript levels of glycolytic genes revealed, that no significant 294

downregulation under the investigated conditions occurred. 295

Glycerol kinase YlGut1 shows the strongest expression among glycerol metabolic genes: Next, we investigated 296

the expression of genes related to glycerol metabolism in Y. lipolytica. As reviewed by Klein et al. (2017), two 297

pathways exist for the metabolization of this compound in yeasts (Figure 4): the phosphorylative glycerol-3-298

phosphate pathway (G3P pathway) and the oxidative dihydroxyacetone pathway (DHA pathway). Both pathways 299

can undergo two directions, depending on whether glycerol is used as a carbon source (catabolic route) or is 300

synthesized to fulfil cellular functions (anabolic route). Glycerol metabolism has been investigated extensively in 301

S. cerevisiae (Klein et al. 2017), and genes from this species were used to identify corresponding homologs in 302

Y. lipolytica (Table S4). It is generally accepted that Y. lipolytica uses the glycerol-3-phosphate (G3P) pathway for 303

glycerol consumption. As in S. cerevisiae, Y. lipolytica possesses one gene coding for glycerol kinase (YlGut1, 304

YALI1_F00654g) and mitochondrial G3P dehydrogenase (YlGut2, YALI1_B18499g). Differences exist in the reverse 305

enzymatic steps since only one cytosolic G3P dehydrogenase homolog (YlGPD1) can be found in Y. lipolytica 306

compared to two isogenes in S. cerevisiae (GPD1/GPD2). The cytosolic and mitochondrial G3P dehydrogenase 307

isoforms are also participating to the so-called glycerol-3-phosphate shuttle (Dulermo and Nicaud 2011). 308

Furthermore, no glycerol-3-phosphatase (GPP) homolog could be identified in Y. lipolytica whereas S. cerevisiae 309

again has two isogenes (GPP1/GPP2). An investigation by Mori et al. 2013 showed that YlGUT1 and 310

YlGUT1/YlGUT2 mutants of Y. lipolytica were strongly impaired in growth on glycerol, but, a slight growth was 311

still observable. The authors speculated that the faint growth could rely on an active catabolic DHA pathway. 312

102

However, designating related genes by in silico predictions is difficult since the functions of related proteins often 313

remain unknown. Even in S. cerevisiae the presence of an DHA pathway is still debated today (Klein et al. 2017): 314

The strongest evidence was the detection of significant dihydroxyacetone kinase (DAK) activity and the 315

subsequent identification of corresponding genes (DAK1, DAK2). Y. lipolytica possess three homologs of the 316

dihydroxyacetone kinase. However, no in vitro activity of the glycerol dehydrogenase (first pathway step) has 317

ever been measured in S. cerevisiae (Klein et al. 2017), while it was speculated that the genes GCY1, YPR1, ARA1 318

or GRE3 could catalyze this reaction (Machiko, Kumio, and Inoue 2004). Interestingly, homology searches in 319

Y. lipolytica result in various homologs to these proteins (compare Table S4). Dulermo and Nicaud (2011) 320

suggested these genes encode glycerol dehydrogenases participating in the DHA pathway. 321

322

323

324

325

103

326

Figure 4. Glycerol catabolic (red) and anabolic (blue) pathways in yeasts. (A): G3P pathway. The catabolic G3P pathway starts 327

with the phosphorylation of glycerol to glycerol-3-phosphate by the enzyme glycerol kinase (EC 2.7.1.30) followed by the 328

oxidation to dihydroxyacetonephosphate (DHAP) by the mitochondrial membrane-bound enzyme glycerol-3-phosphate 329

dehydrogenase (EC 1.1.5.3). As an intermediate of glycolysis/gluconeogenesis, DHAP enters the central carbon metabolism. 330

In the anabolic G3P pathway, DHAP gets reduced to G3P by a cytosolic G3P dehydrogenase (EC 1.1.1.8/1.1.1.94) followed 331

by the dephosphorylation of G3P to glycerol, catalyzed by the enzyme Glycerol-3-phosphatase (EC 3.1.3.21). (B): DHA 332

pathway. The catabolic DHA pathway starts with the oxidation of glycerol to DHA by an NAD+-dependent glycerol 333

dehydrogenase (EC 1.1.1.6) followed by a phosphorylation of DHA to DHAP by the dihydroxyacetone kinase (EC 2.7.1.29). 334

In the anabolic DHA pathway DHAP is dephosphorylated to DHA by a so far uncharacterized sugar phosphatase (EC 3.1.3.23). 335

DHA is subsequently reduced to glycerol by a NADP+-dependent glycerol dehydrogenase (GDH, EC 1.1.1.156). Confirmed 336

S. cerevisiae genes are shown in italic. 337

338

104

In order to obtain a comprehensive picture of the active genes related to glycerol metabolism, we compared 339

respective transcript levels (Figure 3B). During the applied conditions, glycerol kinase YlGut1 showed the 340

strongest expression in both of the strains, while levels were nearly double in W29 compared to IBT. YlGut1 is 341

furthermore strongly induced in the presence of glycerol in both strains. Expression of the glycerol-3-phosphate 342

dehydrogenase, the second pathway step encoded by YlGut2 (YALI1_B18499g), was significantly lower and levels 343

were similar in W29 and IBT 446. An inductive effect by glycerol was observable, however, much weaker 344

compared to YlGut1. Expression levels of YlGPD1 (YALI1_B04433g) were even lower, with again similar values 345

between both strains but no difference between the conditions. Expression of genes putatively related to the 346

DHA pathway was detected. Two homologs of DAK were expressed constitutively (YALI1_F12917g, 347

YALI1_E24532g), however, the levels of the latter were very low. Three putative glycerol dehydrogenase 348

homologs (YALI1_F24773g, YALI1_D09870g, and YALI1_C18771g) were expressed, and levels of YALI1_F24773g 349

were in the same magnitude as of YlGut2. The expression of this gene was higher in IBT 446 where it was also 350

responsive to glycerol. The GRE3 homolog (YALI1_D09870g) exhibited a similar expression pattern and the ARA1 351

homolog (YALI1_C18771g) was only slightly expressed with similar expression levels throughout all conditions. 352

In summary, the transcriptome data confirmed prior studies suggesting glycerol catabolism is mediated by the 353

G3P pathway in Y. lipolytica (Makri, Fakas, and Aggelis 2010; Dulermo and Nicaud 2011). Expression levels of 354

YlGut1 were significantly higher in W29 compared to IBT 446, which is potentially related to the higher 355

respiration rate in chemostat experiments. To the best of our knowledge, it has not been verified that Y. lipolytica 356

contains an active DHA pathway. Indeed, several genes in its genome shows similarities with glycerol 357

dehydrogenases. Nevertheless, these proteins need further functional characterization. As mentioned above, 358

(Dulermo and Nicaud 2011) classified the homologs to S. cerevisiae GCY1, YPR1, ARA1 or GRE3 as glycerol 359

dehydrogenases. For some of these proteins it could be shown that they have functions other than the oxidation 360

of glycerol. It was reported for instance that YALI1_F24773g encodes an erythrose reductase, involved in 361

105

erythritol biosynthesis (Janek et al. 2017), and YALI1_D09870g is a xylose reductase (Ryu, Hipp, and Trinh 2015). 362

The putative glycerol dehydrogenases belong to the aldo-keto reductase (AKR) superfamily. These enzymes have 363

diverse functions in metabolism and the physiological role is often unknown (Ellis 2002). 364

Explorative transcriptome analysis: Using a hypothesis driven approach to detect global 365

expression differences 366

Glycerol repressive effects on genes related to glucose transport and catabolism could not be detected in the 367

presented targeted analysis. To rather investigate global changes in transcriptional activity, we proceeded with 368

a hypothesis driven explorative approach. The conducted RNA-sequencing experiment represents a factorial 369

design with the factors strain (W29, IBT 446) and condition (glucose, mix, glycerol). Principal component analysis 370

(PCA) revealed that most of the variance between the samples can be attributed to strain differences (Figure 371

5A), while the growth condition had a minor influence (Figure 5B). Furthermore, the response to the growth 372

conditions occurred to be ordinal, largely following a linear trend (monotonic increase or decrease of glucose-373

mix-glycerol). The replicates IBT.Mix.1 and W29.Glucose.3 did not cluster together with the other samples and 374

were excluded as outliers from the further analysis. 375

106

376

377

Figure 5. Singular Value Decomposition (SVD) plots showing separation by strains in dimension 1 versus 378

dimension 2 (A), and separation by growth condition in dimension 3 versus dimension 2 (B). Numbers indicate 379

the replicate within the strain and condition group. Dimensions 1-3 accounted for 47%, 14% and 10% of the total 380

variance, respectively, in the RNAseq data set. 381

Almost 15% of genes vary expression level between W29 and IBT 446: To extract the strain effect (Figure 5 A), 382

a linear model was applied as detailed in the Materials & Methods section, resulting in 1081 significantly 383

differentially expressed genes (adj. p-value < 0.05, |logFC| >= 1). Resulting coefficients of the linear model fit are 384

provided in Table S5. The differentially expressed genes were symmetrically distributed with approximately the 385

same amount of up- and downregulated genes (553 and 528, respectively). 386

A gene-set analysis (GSA) was performed to facilitate the biological interpretation of affected differentially 387

expressed genes (Table 3). Several processes were enriched, however, no coherent picture could be drawn from 388

the results indicating mechanisms behind the observed physiological differences: Several processes related to 389

the translation machinery (including rRNA-, tRNA processing and ribosome biogenesis), oxidation-reduction 390

107

processes and transport were lower expressed in W29, while fatty acid metabolism as well as genes related to 391

signal transduction and transcriptional regulation were higher expressed compared to IBT 446. Since the 392

biological interpretation was complicated by the large number of differentially expressed genes affected by the 393

strain differences we decided to investigate strain-specific responses to the carbon sources by cross-394

comparisons. 395

Table 3. Results of the gene-set analysis for the factor strain. Gene-sets have been manually curated to reduce redundancy 396

and only gene sets with p-value < 0.05 are shown. The amount of significant genes (p-value < 0.05) in a gene-set are provided 397

together with the total amount of genes in the gene-set. Blue: gene sets containing mainly upregulated genes. Red: gene 398

sets containing mainly downregulated genes. 399

Gene-set Gene-set p-value sig. genes GO term

acyl-CoA dehydrogenase activity 1.00E-04 11 / 11 GO:0003995

1-phosphatidylinositol binding 1.45E-04 6 / 6 GO:0005545

signal transduction 1.50E-04 44 / 72 GO:0007165

fatty acid beta-oxidation 1.93E-04 8 / 8 GO:0006635

DNA binding 2.00E-04 210 / 297 GO:0003677

protein heterodimerization activity 2.00E-04 20 / 28 GO:0046982

small GTPase mediated signal transduction 2.50E-04 31 / 44 GO:0007264

regulation of transcription, DNA-templated 5.50E-04 167 / 245 GO:0006355

nucleosome assembly 9.00E-04 15 / 18 GO:0006334

aminopeptidase activity 3.75E-03 10 / 12 GO:0004177

oxidoreductase activity1 6.10E-03 9 / 12 GO:0016712

RNA binding 1.00E-04 121 / 177 GO:0003723

rRNA processing 1.00E-04 59 / 70 GO:0006364

tRNA processing 1.00E-04 36 / 47 GO:0008033

translation 1.50E-04 72 / 167 GO:0006412

oxidation-reduction process 2.88E-04 279 / 411 GO:0055114

ATP-dependent helicase activity 3.50E-04 42 / 49 GO:0008026

copper ion binding 1.00E-03 13 / 21 GO:0005507

cell adhesion 4.47E-03 6 / 10 GO:0007155

transmembrane transport 4.14E-02 211 / 308 GO:0055085

1 gene-set name abbreviated

400

401

108

Cross-comparisons revealed that strain-specific carbon source responses undergo in the opposite direction: 402

Cross-comparisons between samples of the same strain were carried out in order to investigate strain-specific 403

responses to the applied carbon conditions. A linear model was applied as detailed in the Materials & Methods 404

section and coefficients are provided in Table S6. The number of regulated genes differed between the two 405

strains, while unexpectedly the strain-specific carbon response regulated genes in opposite directions (Figure 6 406

A). As anticipated, the largest effect on differential gene expression in both strains was observed by comparing 407

the two single carbon conditions glycerol and glucose: In strain IBT 446, 94 genes were differentially expressed 408

with the majority being upregulated, while W29, 61 genes changed significantly under these conditions with the 409

majority being downregulated. Among these genes, only five genes are shared between the strains (Figure 6 B). 410

411

412

Figure 6. Cross comparisons between different samples. (A) Significantly up and down regulated genes for each contrast. 413

(B) Intersection of significant genes in IBT 446 and W29 by comparing the single carbon conditions glycerol and glucose. 414

415

The comparison between glycerol and mixed condition resulted in the lowest number of differentially expressed 416

genes, with only 11 genes significantly affected in W29 (from which six were also present in the glycerol vs 417

glucose contrast) and no significant genes in IBT 446. This signifies that the presence of glycerol in the mixed 418

109

condition is dominant over the presence of glucose. Accordingly, the comparison between glucose and mixed 419

conditions resulted in 15 significantly, differentially expressed genes in IBT 446 (from which 14 were also found 420

in the glycerol versus glucose comparison) and 24 genes in W29 (from which 23 were also in glycerol versus 421

glucose comparison). 422

Gene-set analyses for the glycerol versus glucose comparisons have been carried out. The analysis indicated that 423

the presence of glycerol upregulates various processes related to nutrient scavenging in IBT 446, including the 424

production of exoenzymes (proteases, lipases and glucosidases), transporters and oxidation-reduction processes 425

(Table 4). Processes related to the gene expression machinery and DNA repair mechanisms were negatively 426

affected. Contrary, the presence of glycerol seemed to downregulate lipases, proteases and oxidation-reduction 427

processes in W29, revealing even similar processes have opposite responses in each strain (Table 5). Meanwhile, 428

mainly processes related to stress (starvation and filamentous growth), amino acid biosynthesis and 429

transcriptional regulation were upregulated. The direct comparisons within a strain revealed that in both strains 430

not only cellular and metabolic processes were affected when grown on glycerol compared to glucose but also 431

regulation was involved. Therefore, the next attempt was to investigate if there were significantly differences in 432

the strains specific regulation. 433

434

435

436

437

438

439

110

Table 4. Results of the gene-set analysis for the contrast IBT.glycerol versus IBT.glucose. Gene-sets have been manually 440

curated to reduce redundancy and only gene sets with p-value < 0.05 are shown. The amount of significant genes (p-value 441

< 0.05) in a gene-set are provided together with the total amount of genes in the gene-set. Blue: gene sets containing mainly 442

upregulated genes. Red: gene sets containing mainly downregulated genes. 443

Gene-set Gene-set p-value sig. genes GO term

amino acid transmembrane transport 1.00E-04 5 / 30 GO:0003333

carbohydrate metabolic process 1.00E-04 8 / 69 GO:0005975

oxidation-reduction process 1.00E-04 27 / 411 GO:0055114

proteolysis 1.00E-04 10 / 176 GO:0006508

pyridoxal phosphate binding 1.00E-04 10 / 51 GO:0030170

sequence-specific DNA binding RNA polymerase II transcription factor activity

1.00E-04 9 / 63 GO:0000981

transmembrane transport 1.00E-04 27 / 308 GO:0055085

transport 1.00E-04 20 / 332 GO:0006810

glycerolipid metabolic process 1.44E-04 4 / 45 GO:0046486

catalytic activity 1.50E-04 27 / 357 GO:0003824

cell cycle 1.00E-04 5 / 59 GO:0007049

cellular response to DNA damage stimulus 1.00E-04 4 / 37 GO:0006974

DNA binding 1.00E-04 28 / 297 GO:0003677

DNA repair 1.00E-04 6 / 92 GO:0006281

helicase activity 1.00E-04 6 / 84 GO:0004386

methylation 1.00E-04 5 / 88 GO:0032259

nucleic acid binding 1.00E-04 26 / 308 GO:0003676

protein heterodimerization activity 1.00E-04 4 / 28 GO:0046982

RNA binding 1.00E-04 8 / 177 GO:0003723

RNA splicing 1.00E-04 6 / 38 GO:0008380

damaged DNA binding 2.00E-04 4 / 14 GO:0003684

444

445

446

447

448

449

450

111

Table 5. Results of the gene-set analysis for the contrast W29.glycerol vs W29.glucose. Gene-sets have been manually 451

curated to reduce redundancy and only gene sets with p-value < 0.05 are shown. The amount of significant genes (p-value 452

< 0.05) in a gene-set are provided together with the total amount of genes in the gene-set. Blue: gene sets containing mainly 453

upregulated genes. Red: gene sets containing mainly downregulated genes. 454

Gene-set Gene-set p-value sig. genes GO term

ATP catabolic process 1.00E-04 2 / 98 GO:0006200

carbohydrate metabolic process 1.00E-04 4 / 69 GO:0005975

cellular amino acid biosynthetic process 1.00E-04 3 / 34 GO:0008652

filamentous growth of a population of unicellular organisms in response to starvation

1.00E-04 2 / 63 GO:0036170

lysine biosynthetic process 1.00E-04 2 / 8 GO:0009085

phospholipid binding 1.00E-04 2 / 25 GO:0005543

regulation of transcription from RNA polymerase II promoter

1.00E-04 8 / 104 GO:0006357

regulation of transcription, DNA-templated 1.00E-04 11 / 245 GO:0006355

zinc ion binding 1.00E-04 9 / 376 GO:0008270

triglyceride lipase activity 1.50E-04 2 / 21 GO:0004806

N-acetyltransferase activity 2.00E-04 3 / 29 GO:0008080

peptidase activity 6.00E-03 6 / 113 GO:0008233

glucose transport 1.25E-02 1 / 6 GO:0015758

oxidation-reduction process 1.34E-02 15 / 411 GO:0055114

455

456

457

458

459

460

461

462

112

Hypothesis-driven analysis highlights the involvement of regulatory proteins: Cross-comparisons above 463

indicated that IBT 446 and W29 can have opposite response to nutrients. In an attempt to compare the strain-464

specific nutrient responses, we analyzed all samples together. Three hypotheses were formulated and linear 465

models applied accordingly the Materials & Methods section and Figure 7A: We postulated that both strains 466

possess genes which are responsive to the presence of glucose (Hypothesis 1), while other genes are responsive 467

to the presence of glycerol (Hypothesis 2). As such, these two hypothesis focus on the most conserved response 468

to nutrients, corresponding to Figure 5, panel B. A third hypothesis was formulated to extract genes differently 469

regulated in IBT 446 and W29, where the expression in the mixed condition is in reverse between the strains 470

(Hypothesis 3). 471

By discarding the strain effect from the linear model (which results in high numbers of differentially expressed 472

genes as shown above), the condition effect as defined in the three hypothesis appears to be rather small (Figure 473

7B). In total, ten genes were responsive to glucose in both strains (H1), 18 genes are responsive to glycerol (H2) 474

and 13 genes respond differently in IBT 446 and W29 (H3). Coefficients of the linear model fit are provided in 475

Table S7. 476

113

477

Figure 7. An approach for the investigation of differences in the carbon source response. (A): Illustration of the three models 478

used to investigate conditional effects. (B): Amount of differentially expressed genes according the hypothesis tests related 479

to models 1-3. (C): An example of a gene behaving as predicted by model 3: The expression level of YALI1_A11439g in 480

IBT.Mix is similar to glycerol, whereas W29.Mix is similar to glucose. Expression levels of all significant hypothesis 3 genes 481

are provided in Figure S1. 482

H3 represents the earlier defined hypothesis that a different gene regulation exists in IBT 446 and W29 (Figure 483

7C). Indeed, as shown in Table 6, several of the resulting genes are putatively related to transcriptional regulation 484

(YALI1_A12929g, YALI1_A16891g) or signal transduction (YALI1_E01904g, YALI1_D22368g, YALI1_E14489g). Four 485

114

of the genes are of unknown functions (YALI1_E24676g, YALI1_C13910g, YALI1_F38013g, and YALI1_C10173g) 486

while two are mitochondrial genes (cob: cytochrome B, nad5: NADH-ubiquinone oxidoreductase chain 5). 487

488

115

Table 6. Significantly differentially expressed genes resulting from hypothesis 3. Shown are the logFC and p-value from the linear model fit. Additionally, a 489

blastp homology search has been conducted to receive orthologous protein functions. H3 up genes are highlighted in blue whereas H3 down genes are 490

highlighted in red. 491

YALI1 ID YALI0 ID logFC adj. p-Value E value Identities Description

YALI1_A12929g YALI0A12925g 3.66 0.002 2.00E-18 52/117 (44%) similar to S. cerevisiae YGR044C RME1 Zinc finger protein involved in control of meiosis

YALI1_E01904g YALI0E01364g 2.19 0.004 6.00E-82 148/393 (38%) similar to S. cerevisiae YOR212W STE4 G protein beta subunit, forms a dimer with Ste18p to activate the mating signaling pathway

YALI1_D22368g YALI0D18018g 1.78 0.004 2.00E-25 123/467 (26%) similar to S. cerevisiae SST2 (YLR452C) involved in desensitization to alpha-factor pheromone

YALI1_E24676g YALI0E20779g 1.57 0.034 NA NA no similarities

YALI1_E14489g YALI0E11627g 1.55 0.004 2.00E-119 192/441 (44%) similar to S. stipitis CBS 6054 guanine nucleotide-binding protein alpha subunit

YALI1_A16891g YALI0A16841g 1.51 0.006 3.00E-88 131/243 (54%) similar to S. cerevisiae YOR113W AZF1 Zinc-finger transcription factor

YALI1_A11439g YALI0A11473g 1.29 0.003 0.0 692/1266 (55%) similar to S. cerevisiae YKL209C STE6 Plasma membrane ATP-binding cassette (ABC) transporter required for the export of a-factor

YALI1_C13910g NA 1.04 0.005 NA NA no similarities

YALI1_F38013g YALI0F30437g 1.03 0.023 NA NA no similarities

YALI1_E26094g YALI0E22088g -1.10 0.018 1.00E-18 50/107 (47%) similar to S. cerevisiae YER011W TIR1 Cell wall mannoprotein of the Srp1p/Tip1p family of serine-alanine-rich proteins

YALI1_C10173g NA -1.29 0.036 NA NA no similarities

nad5 NA -1.30 0.034 0.0 568/655 (87%) C. phangngaensis NADH:ubiquinone oxidoreductase (mitochondrial gene)

cob NA -2.81 0.013 0.0 264/385 (69%) K. marxianus cytochrome b subunit of the bc complex (mitochondrial gene)

492

116

Expression profiles of genes encoding regulatory proteins are shown in Figure S1. Interestingly, several of these 493

genes were higher expressed in W29.glycerol: YALI1_A12929g has similarity to S. cerevisiae Zinc finger protein 494

RME1 (YGR044C). This gene is not expressed in IBT, and in W29 expression under glycerol is significantly higher 495

than under glucose and mixed conditions. In S. cerevisiae RME1 is a nucleic-acid-binding protein that acts as a 496

negative regulator of meiosis in haploid cells (a, α) but is repressed in diploid (a/α) cells (Covitz, Herskowitz, and 497

Mitchell 1991). YALI1_E01904g shows a similar expression pattern, but is also expressed in IBT 446. Expression 498

on mixed conditions is in reverse between the strains. This gene has similarity with S. cerevisiae STE4 (YOR212w) 499

a GTP-binding protein subunit involved in pheromone-dependent signal transduction. As a part of a G protein 500

heterodimer (Gβγ), Ste4p plays a critical role in the activation of several effector proteins (Henrik G. Dohlman 501

and Thorner 2001). The homolog of another significant gene is also involved in the pheromone pathway: 502

YALI1_D22368g, shows similarity to S. cerevisiae SST2 (YLR452C). It is only expressed in W29 grown on glycerol. 503

In S. cerevisiae SST2 is a member of the regulator of G protein signaling (RGS) family and negatively regulates 504

pheromone response by stimulating GTP hydrolysis of the activated G protein α subunit (Gpa1p) (Apanovitch et 505

al. 1998; H G Dohlman et al. 1996; Chan and Otte 1982). YALI1_E14489g exhibits an expression pattern similar 506

to YALI1_E01904g and is homolog to a guanine nucleotide-binding protein alpha subunit in S. stipites and other 507

species. It is also homolog to S. cerevisiae GPA2 (YER020W) which is part of a glucose sensing system together 508

with the G-protein coupled receptor (GPCR) Gpr1 (Busti et al. 2010). YALI1_A11439g is low expressed in both 509

strains with the highest levels in W29 on glycerol. This gene is similar to S. cerevisiae STE6 (YKL209c) an ATP-510

binding cassette (ABC) transporter protein, which mediates the export of the a-factor mating pheromone in 511

MATa cells (Michaelis and Barrowman 2012). YALI1_A16891g is weakly similar to S. cerevisiae AZF1 (YOR113w) 512

an asparagine-rich zinc finger protein. It is expressed in both strains with the highest expression under glycerol. 513

IBT 446 shows a linear increase of the expression from glucose to glycerol, while in W29 glucose and mixed 514

conditions have similar expression levels. In S. cerevisiae AZF1 encodes a transcription factor responding to the 515

117

specific carbon source present. Under glucose, genes involved in growth and carbon metabolism are activated, 516

while the cell wall integrity is regulated in the presence of non-fermentable carbon sources (Slattery, Liko, and 517

Heideman 2006). Lastly YALI1_E26094g is also differently regulated in the two strains. It is upregulated in IBT 446 518

on glycerol but downregulated in W29 under this condition. This gene is homolog to TIR1 in S. cerevisiae which 519

encodes a cell wall mannoprotein expressed under anaerobic conditions. It should be remembered, that even 520

though the mentioned genes have similarities with regulatory proteins of other species, it doesn’t mean that 521

their function is conserved. Further research is necessary to reveal the biological function of this proteins in 522

Y. lipolytica. 523

Conclusion 524

Y. lipolytica exhibits remarkable growth capabilities on glycerol, however, most of the current knowledge 525

concerning uptake, catabolism and regulation is derived from S. cerevisiae, a yeast with naturally limited abilities 526

to utilize this substrate. Y. lipolytica differs in several aspects from S. cerevisiae, but especially carbon source 527

regulation is dissimilar. In contrast to regulatory mechanisms enabling the prioritized use of glucose, this non-528

conventional yeast prefers glycerol in co-substrate cultivations. This study embarks on investigating not 529

previously described carbon regulation in Y. lipolytica, by comparing the transcriptomes of strains differing in 530

their substrate utilization phenotypes. Growth profiling demonstrated a strain-dependent physiology under 531

glycerol-glucose mixed conditions. Interestingly, transcriptome analysis revealed the majority of differentially 532

expressed genes between the strains are regardless of the carbon source applied and no direct glycerol 533

repression was observed for genes related to glucose uptake and catabolism in IBT 446. However, several genes 534

were generally higher expressed in W29 including the transporters YALI1_F00616g (YlFPS1), YALI1_C08523g 535

(YHT1), YALI1_D00376g, YALI1_F24031g and glycerol kinase YALI1_F00654g (YlGUT1). Different expression levels 536

of these genes are potentially related to the observed physiological differences and have to be further 537

investigated in future studies. Even though no direct glycerol repression on genes related glucose degradation 538

118

was detected, it is feasible that such effects would be more prominent in different experimental designs. 539

Previous results indicating the suppression of glucose in the presence of glycerol were obtained from batch 540

cultivations, where high residual substrate concentrations can persistently induce relevant signalling pathways. 541

In contrast, the expression profile data here was obtained from carbon limited chemostats where the substrate 542

concentrations at all steady-state conditions were 0 g L-1. Nonetheless, cross comparisons did indicate 543

transcriptional responses to the use of either carbon source during chemostat cultivations, while the genes 544

affected in IBT 446 and W29 were mostly different and their regulation was predominantly in opposite directions. 545

This is signifying that regulation related to carbon source preference can also be observed in carbon limited 546

chemostat cultivations. The analysis of differences in the strain-specific carbon response revealed that several 547

genes related to transcription factors and signal transduction are differently expressed between the strains. 548

Homologs of these genes are involved in the mating pathway and carbon source regulation in S. cerevisiae. As 549

such, this study lays the foundation for further investigations on carbon source regulation and glycerol 550

repression-like effects in Y. lipolytica. Future work should include gene expression studies under batch conditions 551

and elucidate the roles if identified regulators. 552

553

Acknowledgement 554

Patrice Lubuta was supported by a PhD stipend from the Technical University of Denmark. The authors 555

acknowledge the Fermentation and Metabolomics Platform at the Technical University of Denmark for providing 556

access to fermentation and analytical equipment and for the technical support of Tina Johansen, Martin Nielsen, 557

Alexander Rosenkjaer and Andreas H. R. Heidemann. We further thank DTU DMAC and especially Marlene 558

Danner Dalgaard for helping to prepare the RNAseq library. We also thank the students Lars Vindfeldt and 559

Rasmus Thor Nielsen for their assistance to conduct the chemostat cultivations. 560

119

Literature cited 561

Andrews, Simon. 2010. “FastQC: A Quality Control Tool for High Throughput Sequence Data.” Available Online 562

at: http://Www.Bioinformatics.Babraham.Ac.Uk/Projects/Fastqc. 563

Apanovitch, Donald M., Kevin C. Slep, Paul B. Sigler, and Henrik G. Dohlman. 1998. “Sst2 Is a GTPase-Activating 564

Protein for Gpa1: Purification and Characterization of a Cognate RGS−Gα Protein Pair in Yeast.” 565

Biochemistry 37 (14): 4815–22. doi:10.1021/bi9729965. 566

Bolger, Anthony M, Marc Lohse, and Bjoern Usadel. 2014. “Trimmomatic: A Flexible Trimmer for Illumina 567

Sequence Data.” Bioinformatics (Oxford, England) 30 (15). Oxford University Press: 2114–20. 568

doi:10.1093/bioinformatics/btu170. 569

Brückner, Reinhold, and Fritz Titgemeyer. 2006. “Carbon Catabolite Repression in Bacteria: Choice of the 570

Carbon Source and Autoregulatory Limitation of Sugar Utilization.” FEMS Microbiology Letters 209 (2). 571

John Wiley & Sons, Ltd (10.1111): 141–48. doi:10.1111/j.1574-6968.2002.tb11123.x. 572

Busti, Stefano, Paola Coccetti, Lilia Alberghina, and Marco Vanoni. 2010. “Glucose Signaling-Mediated 573

Coordination of Cell Growth and Cell Cycle in Saccharomyces Cerevisiae.” Sensors (Basel, Switzerland) 10 574

(6). Multidisciplinary Digital Publishing Institute (MDPI): 6195–6240. doi:10.3390/s100606195. 575

Chan, R K, and C A Otte. 1982. “Isolation and Genetic Analysis of Saccharomyces Cerevisiae Mutants 576

Supersensitive to G1 Arrest by a Factor and Alpha Factor Pheromones.” Molecular and Cellular Biology 2 577

(1): 11–20. 578

Conesa, A., S. Gotz, J. M. Garcia-Gomez, J. Terol, M. Talon, and M. Robles. 2005. “Blast2GO: A Universal Tool for 579

Annotation, Visualization and Analysis in Functional Genomics Research.” Bioinformatics 21 (18). Oxford 580

University Press: 3674–76. doi:10.1093/bioinformatics/bti610. 581

Covitz, P A, I Herskowitz, and A P Mitchell. 1991. “The Yeast RME1 Gene Encodes a Putative Zinc Finger Protein 582

That Is Directly Repressed by A1-Alpha 2.” Genes & Development 5 (11): 1982–89. 583

120

Dohlman, H G, J Song, D Ma, W E Courchesne, and J Thorner. 1996. “Sst2, a Negative Regulator of Pheromone 584

Signaling in the Yeast Saccharomyces Cerevisiae: Expression, Localization, and Genetic Interaction and 585

Physical Association with Gpa1 (the G-Protein Alpha Subunit).” Molecular and Cellular Biology 16 (9): 586

5194–5209. 587

Dohlman, Henrik G., and Jeremy Thorner. 2001. “Regulation of G Protein–Initiated Signal Transduction in Yeast: 588

Paradigms and Principles.” Annual Review of Biochemistry 70 (1): 703–54. 589

doi:10.1146/annurev.biochem.70.1.703. 590

Dulermo, Thierry, and Jean-Marc Nicaud. 2011. “Involvement of the G3P Shuttle and β-Oxidation Pathway in 591

the Control of TAG Synthesis and Lipid Accumulation in Yarrowia Lipolytica.” Metabolic Engineering 13 (5): 592

482–91. doi:10.1016/j.ymben.2011.05.002. 593

Ellis, Elizabeth M. 2002. “Microbial Aldo-Keto Reductases.” FEMS Microbiology Letters 216 (2): 123–31. 594

doi:10.1111/j.1574-6968.2002.tb11425.x. 595

Ferreira, Célia, Frank van Voorst, António Martins, Luisa Neves, Rui Oliveira, Morten C Kielland-Brandt, Cândida 596

Lucas, and Anders Brandt. 2005. “A Member of the Sugar Transporter Family, Stl1p Is the Glycerol/H+ 597

Symporter in Saccharomyces Cerevisiae.” Molecular Biology of the Cell 16 (4). American Society for Cell 598

Biology: 2068–76. doi:10.1091/mbc.E04-10-0884. 599

Gancedo, J M. 1998. “Yeast Carbon Catabolite Repression.” Microbiology and Molecular Biology Reviews : 600

MMBR 62 (2): 334–61. 601

Grauslund, M, and B Rønnow. 2000. “Carbon Source-Dependent Transcriptional Regulation of the 602

Mitochondrial Glycerol-3-Phosphate Dehydrogenase Gene, GUT2, from Saccharomyces Cerevisiae.” 603

Canadian Journal of Microbiology 46 (12): 1096–1100. 604

Grauslund, Morten, John M. Lopes, and Birgitte Rønnow. 1999. “Expression of GUT1, Which Encodes Glycerol 605

Kinase in Saccharomyces Cerevisiae, Is Controlled by the Positive Regulators Adr1p, Ino2p and Ino4p and 606

121

the Negative Regulator Opi1p in a Carbon Source-Dependent Fashion.” Nucleic Acids Research 27 (22): 607

4391–98. doi:10.1093/nar/27.22.4391. 608

Iida, Toshiya, Akinori Ohta, and Masamichi Takagi. 1998. “Cloning and Characterization of Ann-Alkane-Inducible 609

Cytochrome P450 Gene Essential Forn-Decane Assimilation ByYarrowia Lipolytica.” Yeast 14 (15): 1387–610

97. doi:10.1002 611

Iida, Toshiya, Toru Sumita, Akinori Ohta, and Masamichi Takagi. 2000. “The Cytochrome P450ALK Multigene 612

Family of Ann-Alkane-Assimilating Yeast,Yarrowia Lipolytica: Cloning and Characterization of Genes 613

Coding for New CYP52 Family Members.” Yeast 16 (12): 1077–87. doi:10.1002/1097-614

0061(20000915)16:12<1077::AID-YEA601>3.0.CO;2-K. 615

Janek, Tomasz, Adam Dobrowolski, Anna Biegalska, and Aleksandra M. Mirończuk. 2017. “Characterization of 616

Erythrose Reductase from Yarrowia Lipolytica and Its Influence on Erythritol Synthesis.” Microbial Cell 617

Factories 16 (1). BioMed Central: 118. doi:10.1186/s12934-017-0733-6. 618

Jones, P., D. Binns, H.-Y. Chang, M. Fraser, W. Li, C. McAnulla, H. McWilliam, et al. 2014. “InterProScan 5: 619

Genome-Scale Protein Function Classification.” Bioinformatics 30 (9). Oxford University Press: 1236–40. 620

doi:10.1093/bioinformatics/btu031. 621

Klein, Mathias, Zia-Ul Islam, Peter Boldsen Knudsen, Martina Carrillo, Steve Swinnen, Mhairi Workman, and 622

Elke Nevoigt. 2016. “The Expression of Glycerol Facilitators from Various Yeast Species Improves Growth 623

on Glycerol of Saccharomyces Cerevisiae.” Metabolic Engineering Communications 3 (December): 252–57. 624

doi:10.1016/j.meteno.2016.09.001. 625

Klein, Mathias, Steve Swinnen, Johan M. Thevelein, and Elke Nevoigt. 2017. “Glycerol Metabolism and 626

Transport in Yeast and Fungi: Established Knowledge and Ambiguities.” Environmental Microbiology 19 627

(3): 878–93. doi:10.1111/1462-2920.13617. 628

Law, Charity W, Yunshun Chen, Wei Shi, and Gordon K Smyth. 2014. “Voom: Precision Weights Unlock Linear 629

122

Model Analysis Tools for RNA-Seq Read Counts.” Genome Biology 15 (2): R29. doi:10.1186/gb-2014-15-2-630

r29. 631

Lazar, Zbigniew, Cécile Neuvéglise, Tristan Rossignol, Hugo Devillers, Nicolas Morin, Małgorzata Robak, Jean-632

Marc Nicaud, and Anne-Marie Crutz-Le Coq. 2017. “Characterization of Hexose Transporters in Yarrowia 633

Lipolytica Reveals New Groups of Sugar Porters Involved in Yeast Growth.” Fungal Genetics and Biology 634

100 (March): 1–12. doi:10.1016/j.fgb.2017.01.001. 635

Liao, Yang, Gordon K Smyth, and Wei Shi. 2013. “The Subread Aligner: Fast, Accurate and Scalable Read 636

Mapping by Seed-and-Vote.” Nucleic Acids Research 41 (10). Oxford University Press: e108. 637

doi:10.1093/nar/gkt214. 638

Liu, Hu-Hu, Xiao-Jun Ji, and He Huang. 2015. “Biotechnological Applications of Yarrowia Lipolytica: Past, Present 639

and Future.” Biotechnology Advances 33 (8): 1522–46. doi:10.1016/j.biotechadv.2015.07.010. 640

Machiko, Shingo Izawa, Sato Kumio, and Yokoigawa Yoshiharu Inoue. 2004. “APPLIED MICROBIAL AND CELL 641

PHYSIOLOGY Intracellular Glycerol Influences Resistance to Freeze Stress in Saccharomyces Cerevisiae: 642

Analysis of a Quadruple Mutant in Glycerol Dehydrogenase Genes and Glycerol-Enriched Cells.” Appl 643

Microbiol Biotechnol 66: 108–14. doi:10.1007/s00253-004-1624-4. 644

Magnan, Christophe, James Yu, Ivan Chang, Ethan Jahn, Yuzo Kanomata, Jenny Wu, Michael Zeller, Melanie 645

Oakes, Pierre Baldi, and Suzanne Sandmeyer. 2016. “Sequence Assembly of Yarrowia Lipolytica Strain 646

W29/CLIB89 Shows Transposable Element Diversity.” Edited by Joseph Schacherer. PLOS ONE 11 (9). 647

Public Library of Science: e0162363. doi:10.1371/journal.pone.0162363. 648

Makri, Anna, Stylianos Fakas, and George Aggelis. 2010. “Metabolic Activities of Biotechnological Interest in 649

Yarrowia Lipolytica Grown on Glycerol in Repeated Batch Cultures.” Bioresource Technology 101 (7). 650

Elsevier Ltd: 2351–58. doi:10.1016/j.biortech.2009.11.024. 651

Michaelis, Susan, and Jemima Barrowman. 2012. “Biogenesis of the Saccharomyces Cerevisiae Pheromone A-652

123

Factor, from Yeast Mating to Human Disease.” Microbiology and Molecular Biology Reviews : MMBR 76 653

(3). American Society for Microbiology (ASM): 626–51. doi:10.1128/MMBR.00010-12. 654

Mori, Katsuki, Ryo Iwama, Satoshi Kobayashi, Hiroyuki Horiuchi, Ryouichi Fukuda, and Akinori Ohta. 2013. 655

“Transcriptional Repression by Glycerol of Genes Involved in the Assimilation of N-Alkanes and Fatty Acids 656

in Yeast Yarrowia Lipolytica.” FEMS Yeast Research 13 (2): 233–40. doi:10.1111/1567-1364.12025. 657

Pavlik, P, M Simon, T Schuster, and H Ruis. 1993. “The Glycerol Kinase (GUT1) Gene of Saccharomyces 658

Cerevisiae: Cloning and Characterization.” Current Genetics 24 (1–2): 21–25. 659

Robinson, M. D., D. J. McCarthy, and G. K. Smyth. 2010. “EdgeR: A Bioconductor Package for Differential 660

Expression Analysis of Digital Gene Expression Data.” Bioinformatics 26 (1): 139–40. 661

doi:10.1093/bioinformatics/btp616. 662

Ryu, Seunghyun, Julie Hipp, and Cong T Trinh. 2015. “Activating and Elucidating Metabolism of Complex Sugars 663

in Yarrowia Lipolytica.” Applied and Environmental Microbiology 82 (4): 1334–45. 664

doi:10.1128/AEM.03582-15. 665

Rywińska, Anita, Piotr Juszczyk, Maria Wojtatowicz, Malgorzata Robak, Zbigniew Lazar, Ludwika Tomaszewska, 666

and Waldemar Rymowicz. 2013. “Glycerol as a Promising Substrate for Yarrowia Lipolytica 667

Biotechnological Applications.” Biomass and Bioenergy 48 (January): 148–66. 668

doi:10.1016/j.biombioe.2012.11.021. 669

Slattery, Matthew G., Dritan Liko, and Warren Heideman. 2006. “The Function and Properties of the Azf1 670

Transcriptional Regulator Change with Growth Conditions in Saccharomyces Cerevisiae.” Eukaryotic Cell 5 671

(2): 313–20. doi:10.1128/EC.5.2.313-320.2006. 672

Sprague, G F, and J E Cronan. 1977. “Isolation and Characterization of Saccharomyces Cerevisiae Mutants 673

Defective in Glycerol Catabolism.” Journal of Bacteriology 129 (3): 1335–42. 674

Valerio, Oscar, Tessa Horvath, Christopher Pond, Manjusri Misra, and Amar Mohanty. 2015. “Improved 675

124

Utilization of Crude Glycerol from Biodiesel Industries: Synthesis and Characterization of Sustainable 676

Biobased Polyesters.” Industrial Crops and Products 78 (December). Elsevier: 141–47. 677

doi:10.1016/J.INDCROP.2015.10.019. 678

Väremo, Leif, Jens Nielsen, and Intawat Nookaew. 2013. “Enriching the Gene Set Analysis of Genome-Wide 679

Data by Incorporating Directionality of Gene Expression and Combining Statistical Hypotheses and 680

Methods.” Nucleic Acids Research 41 (8). Oxford University Press: 4378–91. doi:10.1093/nar/gkt111. 681

Wagner, Günter P., Koryu Kin, and Vincent J. Lynch. 2012. “Measurement of MRNA Abundance Using RNA-Seq 682

Data: RPKM Measure Is Inconsistent among Samples.” Theory in Biosciences 131 (4): 281–85. 683

doi:10.1007/s12064-012-0162-3. 684

Workman, Mhairi, Philippe Holt, and Jette Thykaer. 2013. “Comparing Cellular Performance of Yarrowia 685

Lipolytica during Growth on Glucose and Glycerol in Submerged Cultivations.” AMB Express 3 (1). AMB 686

Express: 58. doi:10.1186/2191-0855-3-58. 687

Yuzbasheva, Evgeniya Y., Elizaveta B. Mostova, Natalia I. Andreeva, Tigran V. Yuzbashev, Alexander S. Fedorov, 688

Irina A. Konova, and Sergey P. Sineoky. 2018. “A Metabolic Engineering Strategy for Producing Free Fatty 689

Acids by the Yarrowia Lipolytica Yeast Based on Impairment of Glycerol Metabolism.” Biotechnology and 690

Bioengineering 115 (2): 433–43. doi:10.1002/bit.26402. 691

692

693

694

695

696

125

Supplemental material 697

698

Genome-wide expression analysis of Yarrowia lipolytica strains varying in the 699

utilization of glucose and glycerol 700

Patrice Lubutaa, Mhairi Workmana*, Eduard Kerkhovenb & Christopher T. Workmana 701

702

aDepartment of Biotechnology and Biomedicine, Technical University of Denmark, Kgs. Lyngby, Denmark 703

bDepartment of Biology and Biological Engineering, Systems and Synthetic Biology, Chalmers University of 704

Technology, Gothenburg, Sweden 705

*Present address: Mhairi Workman, Novo Nordisk, Bagsværd, Denmark. 706

707

708

709

710

711

712

713

714

126

Table S1. Putative Y. lipolytica sugar and glycerol transporters. Putative sugar porter genes were identified by 715

Lazar et al., (2017) and a phylogenetic analysis revealed clustering in six groups (Class A-F). The authors suggested 716

that YHT1-6 are the main hexose transporters in Y. lipolytica. Homologs to S. cerevisiae aquaglyceroporins as 717

revealed by blast searches (compare Table S4). YALI1 and YALI0 gene identifiers are provided. 718

719

720

721

722

723

724

725

726

727

728

729

730

731

732

733

734

Yali1 ID Yali0 ID Function Comment*

YALI1_F25587g YALI0F19184 Sugar porter Class A, YHT3

YALI1_C08523g YALI0C06424 Sugar porter Class B, YHT1

YALI1_C12220g YALI0C08943 Sugar porter Class B, YHT2

YALI1_B27645g YALI0B21230 Sugar porter Class C

YALI1_D00094g YALI0D00132 Sugar porter Class C

YALI1_F24031g YALI0F18084 Sugar porter Class C

YALI1_F31381g YALI0F23903 Sugar porter Class C

YALI1_A02335g YALI0A01958 Sugar porter Class D

YALI1_A14024g YALI0A14212 Sugar porter Class D

YALI1_B00357g YALI0B00396 Sugar porter Class D

YALI1_D00376g YALI0D00363 Sugar porter Class D

YALI1_D23885g YALI0D18876 Sugar porter Class D

YALI1_E24245g YALI0E20427 Sugar porter Class D

YALI1_A08672g YALI0A08998 Sugar porter Class E

YALI1_B22321g YALI0B17138 Sugar porter Class E

YALI1_C06222g YALI0C04730 Sugar porter Class E

YALI1_C23601g YALI0C16522 Sugar porter Class E

YALI1_D01234g YALI0D01111 Sugar porter Class E

YALI1_F09965g YALI0F06776 Sugar porter Class E

YALI1_F32841g YALI0F25553 Sugar porter Class E

YALI1_B02283g YALI0B01342 Sugar porter Class F, YHT5

YALI1_B08461g YALI0B06391 Sugar porter Class F, YHT6

YALI1_E27441g YALI0E23287 Sugar porter Class F, YHT4

YALI1_A11557g YALI0A11550 Sugar porter pseudogene

YALI1_C06191g YALI0C04686 Sugar porter pseudogene

YALI1_C14843g YALI0C10560 Sugar porter Pseudogene

YALI1_F00616g YALI0F00462g Aquaglyceroporin FPS1 homolog

YALI1_E06664g YALI0E05665g Aquaglyceroporin FPS1 homolog

127

Table S2. Raw RNAseq counts. 735

This table is large for this document. It will be provided in the related publication. 736

737

Table S3. Transcripts Per Million (TPM) values. 738

This table is large for this document. It will be provided in the related publication. 739

740

741

742

743

744

745

746

747

748

749

750

751

128

Table S4. Y. lipolytica homologs to S.cerevisiae glycerol metabolic genes. 752

753

754

755

756

757

Pathway Function S. cerevisiae

Standard Name

S. cerevisiae

Systematic Name

EC number % identity E value YALI1 ID YALI0 ID

G3P pathway

Glycerol Kinase GUT1 YHL032C 2.7.1.30 43 4.76E-139 YALI1_F00654g YALI0F00484g

G-3-P dehydrogenase (FAD+) GUT2 YIL155C 1.1.5.3 46 0.0 YALI1_B18499g YALI0B13970g

G-3-P dehydrogenase (NAD+) GPD1 / GPD2 YDL022W / YOL059W 1.1.1.8 58 1.85E-156 YALI1_B04433g YALI0B02948g

Glycerol-3-phosphatase GPP1 / GPP2 YIL053W / YER062C 3.1.3.21 NA NA

DHA pathway

Dihydroxyacetone kinase DAK1 / DAK2 YML070W / YFL053W 2.7.1.29 42 3.72E-139 YALI1_F12917g YALI0F09273g

41 2.00E-135 YALI1_E24532g YALI0E20691g

37 2.49E-126 YALI1_F02508g YALI0F01606g

Glycerol Dehydrogenase (NADP+) GCY1/YPR1 YOR120W/YDR368W 1.1.1.156 50 2.16E-91 YALI1_B09211g YALI0B07117g

49 1.56E-90 YALI1_B28394g YALI0B21780g

49 6.50E-90 YALI1_D05111g YALI0D04092g

49 6.93E-90 YALI1_E22041g YALI0E18348g

49 2.70E-89 YALI1_A15863g YALI0A15906g

45 7.92E-73 YALI1_F24773g YALI0F18590g

Arabinose dehydrogenase (NADP+) ARA1 YBR149W 1.1.1.117 44 1.71E-73 YALI1_C18771g YALI0C13508g

41 1.95E-65 YALI1_C12619g YALI0C09119g

43 4.79E-65 YALI1_F10224g YALI0F06974g

Aldose reductase (NAD(P)+) GRE3 YHR104W 1.1.1.21 31 1.56E-33 YALI1_D09870g YALI0D07634g

Glycerol uptake Aquaglyceroporin FPS1 YLL043W 34 0.00 YALI1_F00616g YALI0F00462g

35 0.00 YALI1_E06664g YALI0E05665g

129

Table S5. Coefficients of modelling strain and condition effect 758

This table is large for this document. It will be provided in the related publication. 759

760

Table S6. Coefficients of the cross comparisons. 761

This table is large for this document. It will be provided in the related publication. 762

763

Table S7. Coefficients of the hypothesis 1, hypothesis 2 and hypothesis 3 testing. 764

This table is large for this document. It will be provided in the related publication. 765

766

767

768

769

770

771

772

773

774

130

775

776

Figure S1. Significantly differentially expressed genes according H3. Expression levels are shown in log transcripts per million (logTPM) and names of 777

S. cerevisiae orthologs are provided. 778

779

131

Conclusions and Perspectives The principal goal of this Ph.D. study was to investigate the physiology of Yarrowia lipolytica in order to use

alternative carbon substrates. Expanding the range of feedstocks to be used in industrial fermentations is an

important challenge of the bio-based industry. Y. lipolytica can be isolated from various environments and many

different linages are used currently by laboratories worldwide. Since strain-to-strain variation was observed

previously we initially decided to investigate different Y. lipolytica strains. Two of the strains have been selected

since they are commonly used in the research community. W29 is a sewage water isolate and is also known as

the French strain while H222, the German strain, was isolated from soil. Hyphae formation have been reported

frequently for W29 and H222 complicating cultivations in industrial settings. In contrast, the Danish strain IBT

446 seems to have lost the ability to undergo yeast-to-hyphae transitions making this strain especially interesting

for bioprocess applications.

A benchmark of the three strains demonstrated that Y. lipolytica is highly suited for glycerol-based applications.

Growth rates were higher on glycerol than on glucose, a substrate that is also well utilized by this species.

Glycerol was also the prioritized carbon source in mixed substrate cultivations. It could be shown that Y. lipolytica

produces polyols as a natural product. Polyols are sugar alcohols, which are increasingly used as artificial

sweeteners in many industrial applications. IBT 446 showed the highest polyol yield and further studies should

address process optimization strategies including the use of high osmotic pressure conditions.

While Y. lipolytica is a promising host for glycerol applications, results demonstrated that pentose utilization in

this species is limited. We could show that none of the strains were able to use xylose and arabinose when

applied as the sole carbon source. Interestingly, the three strains were able to use xylose in the presence of

glycerol and glucose. IBT 446 was additionally able to use arabinose. The results suggested the existence of

endogenous pentose pathways in Y. lipolytica. While the investigation was conducted, several studies have been

published reporting the existence of the endogenous xylose and arabinose pathways. However, the expression

of corresponding genes is insufficient. Many research groups aimed, therefore, on the overexpression of native

and heterologous genes in order to improve pentose consumption. Future studies should include genetic

engineering of IBT 446, since this strain appears to be naturally better suited for the utilization of pentoses than

the other two linages.

132

An increasing amount of Y. lipolytica strains have been sequenced currently and the availability of genomic

information facilitates the establishment as a novel cell factory. We sequenced the genomes of the three strains

due to the observed physiological differences. The Y. lipolytica genome contains six nuclear and one

mitochondrial chromosome and consist of 20,500,000 base pairs of DNA. The identification and analysis of single-

nucleotide variants (SNV), revealed that the W29 genome differed only slightly from the reference genome

YALI1. In contrast, both H222 and IBT 446 varied significantly from W29 and YALI1 but showed a higher degree

of similarity with each other. The constructed draft genomes will facilitate implementation of genetic

modifications and allow further analysis of the natural diversity in this species.

An RNAseq-based transcriptomics approach was used to investigate the carbon source regulation in Y. lipolytica.

No specific glycerol repression could be detected in IBT 446 but several transporters were constitutively higher

expressed in W29 potentially explaining the observed physiological differences. While no glycerol repression on

genes related to sugar uptake or catabolism was observed under the evaluated growth conditions, we speculated

that such effects would be more prominent in different experimental designs. Previous results indicating the

preferred use of glycerol and the suppression of glucose were obtained from batch cultivations, where high

residual substrate concentrations can persistently trigger relevant signalling pathways. In contrast, the

expression data were obtained from carbon limited chemostats where the substrate concentrations at steady-

state conditions were close to 0 g L-1. We speculated that the low extracellular substrate concentration affected

the putative glycerol repression system. Future experiments should aim to quantify the gene expression during

different growth phases in batch cultivations or in non-carbon-limited chemostats. Interestingly, reviewing the

literature revealed the existence of a W29 strain with an opposite substrate consumption phenotype than the

W29 strain in our study. This strain exhibited a strong suppression of glucose utilization in the presence of

glycerol and future studies should include both of the W29 variants.

The genome-wide expression data was also used to address open questions related to substrate transport and

catabolism in Y. lipolytica. Compared to S. cerevisiae, much less information is available concerning hexoses and

glycerol transport in this species. There is evidence that glycerol transport in Y. lipolytica is not mediated by a

glycerol/H+ symporter (such as Stl1 in S. cerevisiae) but by a homolog to the aquaglyceroporin Fps1. The RNAseq

data confirmed the importance of YlFPS1 since the expression is strongly induced under glycerol levels exceed

those of the other transporters. Very large differences also exists in sugar uptake between Y. lipolytica and

S. cerevisiae. In S. cerevisiae hexose transport is mediated by a single group of sugar porters (HXT1-17 and GAL2).

In contrast, putative hexose transporters of Y. lipolytica are distributed among several groups, from which the

133

HXT-like transporters play only a minor role. More important is a group of transporters showing similarities to

glucose sensors in S. cerevisiae (Snf3 and Rgt2) and a group with similarities to a high-affinity glucose transporter

in K. lactis (Hgt1). Based on the expression data we could confirm previous indications that Yht1 and Yht4 appear

to be the main hexose transporters in Y. lipolytica. However, it should be considered that the importance of

these transporters apply specifically to the tested conditions (carbon-limited chemostats) and that other

transporters could be important. In S. cerevisiae sugar uptake is highly regulated and the HXT transporters differ

in their substrate affinity. The extra- and intracellular sugar concentration determines expression of the

appropriate transporters. Future studies must aim to decipher the roles of the remaining transport related

proteins in Y. lipolytica and the condition of their expression.

Finally, the existence of a DHA pathway in Y. lipolytica remains unclear. In contrast to S. cerevisiae, the deletions

YlGUT1 and YlGUT1/YlGUT2 do not completely abolish growth on glycerol and it was speculated that the

remaining weak growth phenotype could be due to an active DHA pathway. Three homologs to S. cerevisiae

dihydroxyacetone kinase (DAK) and various homologs to glycerol dehydrogenases are present in the Y. lipolytica

genome. However, the function of these proteins are mostly unknown and physiological roles could be distant

to glycerol oxidation. The putative glycerol dehydrogenases belong to the aldo-keto reductase (AKR) superfamily

with diverse functions in metabolism. For some of the putative glycerol dehydrogenases it was shown that the

substrate affinity is actually higher for other compounds (such as erythrose or xylose). Future biochemical studies

are necessary in order to elucidate the function of these proteins and to answer the question if a DHA pathway

exists in Y. lipolytica.

Taken together, Y. lipolytica is highly suited to be used as a cell factory for the valorization of glycerol, even as

the wild type. This yeast can naturally synthesize many economically interesting products from glycerol and

metabolic engineering can expand the portfolio even more. Future studies must address potential problems

related to the use of raw glycerol. Wild type Y. lipolytica strains are naturally less suited for the use of pentose

sugars. Endogenous pentose degradation pathways exists but genetic engineering is needed to activate these

metabolic routes for the efficient use of lignocellulosic sugars.